Add to Watchlist
Fuzzy retrieval model, Coordination level matching, Vector space retrieval model (13.4.2011)
158 views
Citation of segment
Embed Code
Formal Metadata
Title  Fuzzy retrieval model, Coordination level matching, Vector space retrieval model (13.4.2011) 
Title of Series  Information Retrieval and Web Search Engines (SS 2011) 
Part Number  2 
Number of Parts  13 
Author 
Balke, WolfTilo

Contributors 
Selke, Joachim

License 
CC Attribution  NonCommercial 3.0 Germany: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal and noncommercial purpose as long as the work is attributed to the author in the manner specified by the author or licensor. 
DOI  10.5446/353 
Publisher  Technische Universität Braunschweig, Institut für Informationssysteme 
Release Date  2011 
Language  English 
Producer 
Technische Universität Braunschweig

Production Year  2011 
Production Place  Braunschweig 
Content Metadata
Subject Area  Computer Science 
Abstract  This lecture provides an introduction to the fields of information retrieval and web search. We will discuss how relevant information can be found in very large and mostly unstructured data collections; this is particularly interesting in cases where users cannot provide a clear formulation of their current information need. Web search engines like Google are a typical application of the techniques covered by this course. 
Series
Annotations
Transcript
00:00
Orion Lawlor your body of work today it's just me because it was about his attending a conference in Manila today on their whom only today we will have a look at some long sophisticated feel more than the boiler investigated the said the more than used by most of the state of the art of the 2 systems for example of Astronomy believe that we will do realise on some of these techniques presented to date but the focal commented that I'd like to discuss the homework sides with you so who would did the home where the size of the 2 people you can access cheque what they tried to do it and the other on the way there I learn something new and as said last week it was a good deal to do this although exercises because it's the only chance to get immediate feedback from last before the excesses of otherwise you have to have to learn and or said that apart the excellent and he can just try out some things going it guess is to us and we will get some Details feedback
01:21
A Case 1st to exercise and there are 5 days you linked to the sea and Computing classifications system that while some the this some some mesh like system provided by the CE and to classified research applications from the area of Computer Science so it was last revised in 19 98 so it's not not to walk up to date on the part of the axis size was Holloway a book about the Web such and such as Blue would be classified according to the system so I'm not open on this litigation the website at least try to do so by law
02:19
A pick of tries and then you were Italy where to work
02:33
Classified this wonderful book of only it seemed that we don't and that excess and
02:44
From there it out a carried on the internet was the most fundamental principles of computing will now be looking to add a you to do this kind of some some people take on what you tried so what has been your results where you're your 1st impression don't of the as yet investigation system
03:21
But gag so that Sir Araba broached tulle pick dead
03:44
From here you can have the same computer legacy far that it that the point of a accoucheur right now but that is not a big problem such this education system is rather smaland covers only rather broke topics of any book written about such technology would be classified into a information retrievers something and that part the books and publication to that and I recreated every year so using subject as occasion system is not very helpful for practical purposes when you looking for a Pacific books are publication but it may have librarians assault the shed of some way of this part of the problem is that occasion Systems you must maintained its there must be must be side by sophisticated to be really helpful and the and the and has investigation Systems usually do this did not mean that the part of a much so some some publishers require that you that you can papers according to the systems and then you will use the end up in some way or information systems something wrote that the case about it in the area for because it because from sizes you feel and even information systems rather large of some or all of things we are doing here is in situ to the information systems databases and so in the current distinguished this different parts to a simplification came out a 2nd as exercise are last researchers example of these mesh classifications systems from the Mediterranean and the question was what possible prop the problems with these complex classification schemes and in which scenarios are on these approaches amazed those approaches we would the for any opinions on that that oppressor question on stream from today and will be again in the fall for each year term have after review all documents but usually if you have some the main experts at hand they know where the cell documents had would be classified before because Medline strongly strongly emphasise that has a case patient is consistent some way so if you can Utermann you disease that would have been classified different in the past and you are able to find rather than publications and reclassified them going so you to the right accuracy is is large benefit of these schemes but it would have cost require that you keep this scheme up to date so meshes updated every year and I really really needed highly complicated justification treaty with many many small sub area to be able to to find which looking for and get a complete picture of fuel to research domain in this case medicines some of grass it's a huge amount of work and it usually is very expensive so it's only worth the effort and scenarios where life depends on this knowledge or or a science could not do well without it got no debate of not being able to with with further this specifically to some somewhere that have been published in the past so that was a trade off you have to make desiccation schemes can be helpful but but the we have full they have to be expensive
08:11
It is Yes that's about the point so so because it is handcrafted and designed to to hit the streets of mental where the scientist and people working with it usually have let understand below system presentation that that can can be be used by people who had used searching for information for classifying of cause you need some trained experts to do this consistently to find information duties suffices to to browse through the tree as you do it with a see him example alright sign that exercise or a we have seen intimated who is about to take documents filed in relation a database this for some time now we also could be the stalk text documents for example accurate large object that Tybalt we could unit the Bank of what's Moloto database table with 2 at the beauty of the 1st term of a 2nd is documented name may be as the economy indicating the number of provinces of the city in there in the respected document so why while the heck do indeed information between systems because could use databases but what so that the point about the method you We are were really we really coach coach put the bag of words smaller into a database to fewer would like to 1 of the same colour as the document the atomic rose column is the number of accounts of the to build a new low full it as a 2nd the points a basic early this thinking between databases information systems itself historical off electrical nature of databases try to focus on the expected for Utah's bacon defined Logica conditions on your delicate looking for more information feel for the ball to best at off and red events and such Wakeham's like information need so you try to reset look but you can see intuitively might be looking Fault might be relevant to them and do this and in a most mostly effective and efficient way and based systems on just about of filling up logical constraints in some way on the Effect there are some approach is trying to combine the into 1 big system that we are just at the beginning of this area so some databases currently office appalled takes Search so you could do what he would based exarch on all NHS takes documents Dalton database for example of using a catalogue large object that I've the usually these 2 paradigms databases and made to do not integrate well so they also some intimated Philip Systems trying to use structure data for example in on and shops often on these are found that the dust has browsing on the 1 side was Greenway for example coats could select only on the Web from costing less than now 150 euros or something and he just take their respective cheque books on the downside of the purchase tied a combined those usually that it's not that easy because these a completely different perspective on the problem and the hot integrated or night from next 1 somebody had the meaning of the term information need and red and information treated and 1 the connexion between both the It's the a cake The look is that by So relevant is completely subjected concept the have been Kelly behaviour that it rather than so far too but it differently many many different different ways to define relevant see the and some lectures on this could be done that actually different agrees of rather than some people say events is completely personal thing and even if I'd search engine shows me document that completely unrelated to my initiative create but had to in some way because I'll have some of and some problem and them on private life want to sort but did not create but this time then this could be a pivotal to me is that this will be ready and on the other hand said that people saying that relevant can be objectified in some way that each usually people out she agreed that the document is red event on not and relevance could also be a great great it saying that rather than the sort of and but brillant medium level relevant so dominated conditions to do this summer and information Eden Reddin's allways go together because that events this now what but satisfied this information each from what was a prospective these December a victims and that's 1 of the main problems of information Tree will and but also a big a distinction between the 2 0 2 different area because information feel systems but if you build a new system are designed antilogarithm you are not able to prove rigorously in mathematical terms that this algorithm brings the right results because you don't know what I can to find a strike your see that the only chance you have can compare azides to what you man's Bullwood manually have found red event or have and so it's some kind of empirical work you have to do when you designed to make the trip into a 3 weeks to take a look on not call or information to order rhythms can be very rated in a more less objective fashion but that was a big problem information treated like a fine now the difference between the back of what's small and the set of West model paid as the 1 in 1 out equable said that in the case of instead of small will each document is represented as a set of clothes and in the bag of smaller of its herbs and as a bag of words which is still a way to stay in the same muddy said of that was mounted the province of the same time so sometimes mathematics mathematicians use mighty sets and the fact that really really have information Fuso said it would slow Israeli used beauty use bag of what's model but called the whole off some because undocumented is highly of that about documents continent easy enable you to buy the other half were 1 of the topics of are this lecture from when his movie industry full because the bullying looks like a database like information on the tree order rhythm and not to have full of finding out of 4 for really finding information you looking for a way out of a possible scenarios the bullying fever can be useful and why that
17:29
For serv to the Part of it You need Highlight the necessary reduced the number of low because you can't control the number you can new can use a bit more relaxed fulmination a victory which the with less this constraints and the the had huge results so you really don't know how many results to get control in results size and is the tricky including the tree Ombu yet database like such right but we before tax documents that in the end different from from of debt from a better databases to usually and that the a new from the sort things in domains they usually have a high standard rise to February but that offer occurs technical domains some such on a database of technique specifications of patents for example that this might be helpful off stump in the domain some people say this could be because the quite helpful from from such big advantage of wounded feeling is that you really can very fine but the system returns to you go when you just endocrine you have no idea why exactly the about the results you actually got of the under the skin and even if the algorithm would be would be would be people with published by good old be made available to the user is no chance trying to really reconstruct what has happened in the background all these algorithms to get the the results that she was told you if you really want to talk about what's happening in the industry with the tree with a go is a new 1 to use the web as a set its quite get together a tricky to use it because it was the problem by because controlling size but you know what you doing so well all nothing that the last round move so we have seen in the US lecture when we representing on documents on the go collections McGinty was shown document Matrix of rose off the sun columns of documents of the other way around does not a summer of taste and but do these may traces of the response of the very very automatic smaller number of entries on non 0 because most terms do not occur document undocumented Computer Science and won't have many psychological key in of cost so I'm from my experience usually some 1990 when 9 per cent of of the start of the system document Matrix contains 0 was empty from what it does seem to be a good way to start is completely in memory as long as I was in need of a new imagine any better ways to do it again
21:33
Part of that all In lieu of the system Tree lacked structure of the Trump for some than you would start to or you start with a anti roots and and the and the character of the area you would make a Branch and some ideas that died and from this also again with you and it's not if the greatest finished would be the are and then he would be a reference to the document where this democracy and that the way Suffolk of critics trees where you might like to call it is because I'm a deal but this account of the used to define terms in large lodged so and in the place where they did than that a year or you are able to start a database of every very efficient so you don't need huge amounts that is because when you understand how internet for the whole replica will want to do that can point to the memories and hot this with 0 said nothing and because it is information although estimate to be searched through the they have any real eye on a compact representations sometimes almost lost most popular as it is is representation as it is based firm maybe you are also at some information about how often the turmoil Persea somebody of the 2 would mean in documentary the 1st term Walker twice and document 9 it occurs only wants and then you can some instead structure and easy ask and Luis this sometimes you can even apply compression on goes to these lists because not being a smart amount of debt from a slow disc and during the decompression in memory could be Foster off told and reading large list from from the disc so that are highly highly sophisticated ways to do this we will look at some of them in are moved to a three week and if you are really going for it fishnet field so if you want to make a search engine really really Foster this is the construction site to have to work during the representation 93 on these basic list led a very very efficiently and using a stable you cash all you can do or using the Kirov distribution system in Union the about across many different conditions of so that many ways to do it and is usually get up part designing search engine so the algorithms we are will discussing the lecture usually could be could be easy to be understand about about its its I using the a big trick that implementing them efficiently that 3 all part of doing the building a search engine is 1 per cent having a co idea that some of the important part of calls but it's 1990 sent off the engineering so that told works
25:25
The of has you coach of Fuqua which she was a database for this time but usually you're a new using the city designed it as part of the for this because I don't need transactions in from information systems and bit such index that basic idea how these things are a nice early be trees are sometimes use information trio for matching the data but it's not used in by wide by a pudding or the dedicated eBay's and then running some reason on the database because that would be far too slow the databases optimized Fault modifications of their usually and for some transactions and that a consistency and all the stuff but they are not designed to modify information on access information on Saturday 11 as you want to do it here and I want to make a difference in information to assistant is you really want to control the whole Dennis actually stop early on that basis you can only say that we have got to do with all terms and documented Marion index on it and the database completely decide how to manage all the stuff so that if you don't know what the databases doing and if you can not can out of trick this starts in some way that it had Hewitt efficiency then maybe and of the beauty slow system so databases in principle could do his work but he usually on duty on the part so of that also means bidding information to the system that who will have a scan the basis of a lie means building anything from stretch from scratch you can use any existing not brought of technology because they are usually made 4 different purposes then you need to do in this case of light and so that these questions are very excellent like what outside they saw in the Exxon we don't require you to and some stupid conditions that you read on know what we want to talk about all the concept but using the next all these ideas and be able to critical I reflect them and discussion with and tell us what sort working group of working and why things and on the way they are usually on reasons it is costing the day some clever idea behind it and there is some motivation to about why people are doing it this way and and do it another way if you are able to explain to to us why things done this way and by other solutions would be stupid and that would be a great eczema doing alright money to go on slide that are paid not to the contents of today's lectures some of the 1st week of this lecturer only some of this really discussed and that sizes so we found about 100 to help documents on set of where to next terms reduce are published a fabulous so that and China and mankind something like this results is a set of documents that what it is is exactly the set of the documents satisfying the and folk somebody to documents 3 sets of terms that caused his death mankind meant that China tycoon not and the China mountain year than the previous and China and economic or men you could use that could see it through to a easy state you are on your information in some way and the result would be document 1 and 2 so it justice that because we don't know which documented fits the decree that because those satisfied for this as no grated equated way to Dutch which document have small and that the document is not return to the cost of the work contains so that it doesn't need a contained men not China and taken out of the rather easy a but we've seen on working with the set up and and a by reams membership of terms and documents often is not very high because on the usually the next terms in the document that are more than 40 than out of it I'm writing a documentary about the history of Computer Science and probably the term history would approach somewhat more often often than some name of some something from a scientist in the document and that it would be would be halved if someone is looking for Computer Science history to just to to erase the bulletins of this document because it contains Computer Science very often and history very often appropriate more often than other documented only in that it is time history and the don't designs just because of 1 of the idea was to use so called on the index terms and father means found that you can find waits to the terms in which documents to see examples of and all what your as can do with his weight is you can account for and of documents time so if I'm writing a document talked about science and general on and I'm not using the Charm research and it would be really great if people would be able and that looking Fault faultless for research would be able to find my signs document because beaches and signs of someone created in a way that exactly 7 but they are highly highly related and so it would be a great thing if Sisto automatically was signed in the same way to them synonym slightly lower than to the Richard and is the main motivations of using waited approaches pseudonyms are related terms and number of them is an acute document that left him I do it weekend and improve the bullying and describe documents by so called has set and funded said means that now terms on don't have any do any more they're binaries said Mehmet should membership of the document migrated membership as said some document that occurs in the document which have great 1 and some related from that does not occur document with a rate of 0 point 5 something because it is related but not does not occur in the document not advantage before using weight in some way with able to view the uses waits for computing arraigned list of lorries or so we are highly paid with 2 0 2 0 1 0 results and that the most rather than documents in front of the list for use easy can find of Italy but said that in June to for long time now and we had to do that will not take a close look at this point this other to lower works and then we take a look at 2 other from the popular privileged that similar from their deal off of using using waited terms and find the the air because next week with going to talk about publicity lose to give a Chaudhry off abilities you read because of time assuming that most of you will on off on off of to experienced and not the to all these of 2 through to experience and all these terms and just needed sure fresh on that standard 500 feet so I saw other we have all which undocumented stood by demand of people a climbing and and some mountains in China like miscellanea the famous mountain the and depending on the document it could be the case that it's not about how he steps followed that her but highly climbs somewhere in China and so it would be a great idea to assign highway to China and also highway to mount the and only is that only a low wage to the step in this document so assigning with its looks pretty the follow up to the problem is how to a halt to work now with these this kind of sex to to hold recompilations with its problems on how to compute with the said some way and where do we get this degree agrees from the top of those problems from not so I'm not only panacea on so called the logic of of this is left the not a in the Sixties he had them at the idea that of bullying said membership of a member something is contained in a set or it is not might be a bit too restrictive far event applications and so I decided what possible transvalues are not just in true but also any number between the 2 numbers of some of the strange and actually in the Sixties many people Phlegyas his she's just Gilbert and it did not and the and understand anything of bullion of bullying that actually a time he was exactly right in now people use these concept because when you when you talk about 10 per cent then you usually it that is some kind of great great concept of people were below the with say 1 million and 17 would be considered told or so maybe kindergarten but not in the most of the nose and that of course was where people that doctors have and to media and 10 for example of which definitely be consider and everything in between now that he is rather thought he is quite a lot yeah you really are about his also also told followed than the average person that these are the terms people used in real life and so that is just idea represented a set of 4 per cent rise in the greatest of all for person's but by taking list of where people and assigning tallness caught them in some way again to the next thing is now we know that we have some set with great said membership followed it would be a great thing if I just could translate bullion operate as we learned about last week of 2 would operate on in the face of the set and the sums on design built up the position of Logic with only 0 and 1 should be a special case of Logic found that wants to do it was to develop and fuzzy the operators about half combined of the said should of some mathematically nice but he up that much Shia where they want to know more about can to invade the W to be completed out of order because Wikipedia about where the period that far purposes and this lecture found its enough to know the following year saw that introduced some notation a new have a choice if it is a is some is some great it by the referee and so effect it's a number between 0 and 1 of some of the players degreelevel person than me you that no and them a a is just about it and new of a is cost is a degree of class membership of this area of a is is this time if the membership of the person in the class of 4 sets the new of a would be its membership
38:38
Degree or whose value if it is to transfer to bully Logic so and his ideas consent Lachenmann say a and B so combined to which the to truce values by conjunction than its was value would be the minimum of both individual values So if Someone areas the beach tell and very large then he used to be large and thought of the of or your estate NEMO value so far this junctions about you just take a maximum of growth and foundation simply 1 minus disvalue so as reconceive possession was Logic indeed is that the case because it is that we have a view of the IBM was in this set 0 and 1 the and the minimum of these was value is indeed the truth value of seconds than conjunction and the maximum is the truce value of their disjunction and 1 minor the truth value is exactly the indication of just just taking taking the ideas of bullion Logic step ought to on waited a fuzzy membership agrees to obtain not some that said that the document containing features that China and Mountain the with different membership agrees that the documents is highly rather than to the term China into the terms mountain I and it's just a little bit about steps and anyway and the critical the step up not China all multimedia and the question is how well about rather event and is this document the suspected the free so him result this 0 point and how do we write this this value in ideas and so excited we have not not China would be a new 1 1 at the end because but not and not this would be an a no and his at the minimum this would be 0 point 1 and this is 1 of this was maximum and the result 0 1 8 so far this document is contained with degree near point in the set in the scent of all documents that satisfied these 3 because it satisfied that the mountain Utermann the mountain aeternam it's quite representative for this document disappointed that some of the uses just brilliant bullion logic in bullion retriever taking a step for forward to agrees and if you use this correctly on in a reasonable way and and can end up with bated results at the rank results and so far to the problem is whether this is are really she would just like as set these operate as we have to find has really nice properties global from mathematical perspective who has reached the every about Chenault and she called non that all different ways to define this stuff has said it doesn't matter but when looking at these examples he sometimes on these operators are not doing what we expect them to what we would intuitively expect them to do and founder of the 1st documented has assigned timestep in China those with the power the weight of the point for the 2nd document contains a time step in China where the weather that has assigned to on 3 in China the weight 1 China is really fully contained in the set and then that could be step in China and Far and means taking the move and this would result in document document 1 the of the getting good the relevance degree DuPont and went to the relevance to greet the required 3 because protect us from the moment as to concede China is true is completely contained in the document and that just a little bit less than and this document part of of last time this document is ranked high because step so the minimum as 0 point for and and many moments just taken the most postponement of and thing and so we would expect as China has such a large large weight the document to would be would be more than an document 1 of the limitations of these operators to the field to the other mathematical visions of William logic and some of the series they have to be divided defined in this way and that we get some Scott strange results he found to put in my because this problem around so 0 if you have a query someone a and term to and this would be the membership degrees of someone and term to the where these documents lying on this line would be assigned to the same the same membership degree 0 point 7 because for are these documents the me is due point 7 and the members and and the membership of the 2nd term of office with the 1st of his completely note this rugby by the usually don't want to have here because it was the 2nd time workers very often and this of cost should have some effect on the result so of Simoneti about when you in a new of the look of William or in the creative then or this documents you realise that on the same line again membership degree viewpoint 7 and it doesn't matter how much the other the term is contained in the document are less so I would expect if their changes in this direction Bacall's terms to focus on to is more open and he was document on not and this really should change Dolores got so this kind can be done easy see how the the week so limitations when dealing with the real problem
45:44
But they are scenarios where you really think you suggest You opposite of the Dead I live along the a book a have devastated ostinatos where a well handling handling written in this way might be reasonable but using now is this is not database like of says so I have some law to with hot criteria and but was that we NHS should be satisfied as good as possible and so gap with what you would require from from a database system that is able to cope with great membership of information refuse Systems may be use should rank of information higher and should not be can or the other information about the death of that that then and because of this women during fuzzy fevered usually is used in of like an area and either to popular automated to but it it shows that there are 2 other come from believe a tree with 2 more sophisticated and more a case and problem is where do we get these the membership decrease from because inorder undocumented we only know that document that chemicals undocumented undercut by a man who had been offered a the could use this you do to to attend translated into a into waited agrees but it doesn't solve all set in a probe problem of some of the said that the terms of research and science and a man documents only science occurs that it would be really Brady also assigned the term research to documents may be too early to a degree of the point of this because of the related so assigning membership agrees could be done many of the best nice person I usually is a lot of what we are and computer scientists looking for on Wall automatic solution to this problem and the US now has given some back of letter presentation of documents the question is how to convert it into a some reasonable fuzzy said the presentation because that situation usually have because in advance we don't know what terms we ate it off semantic in some way we just know what terms Walker in or collection of documents so and 1 way to transfer of these document of this time document cantonment into into what heavyweights proposed by about and and some of his colleagues and 91 and the deal was in this case 1st takes the requests set after that occurred which document and assigned somewhere terms to the documented obviated to the terms of current the document so far points out that if we have this document that contains timestep China and Mountain would assign step China and Mountain you all the way it 1 and some created terms should also be added to are decorated resentation parts of the UK and this is related to mount in the UK and Asia is a summary to China are and the way to sums makes clear Hall strong this relationship it's of this book is presentation we would be useful and inquiries and if people are not looking for documents where the finest do something in China we are not able to find this document and but usually with no chance of finding the supplement so is busy idea and not as it seemed that assigning these new terms to the document and and sunny waits to them has something to do with Tom Simoneti so much in the yet same way that the way to a 2 2 at the this and I would like to automatic early derives the simulated he somehow and there is some way to do this please call the Jakarta next which we take a look at not about it said the measures were terms tend to walk are in collection of what used to execute together and documents and a given 2 terms she and you than the checkkiting next with a call to see off the and you and it simply the number of documents that contained but terms divided by the number of times containing entities 1 of these is just the the room but if amount of documents in which was terms of grow given that at least 1 of the terms of to are complete synonyms and are always appeal together in a collection and they would have a tough time index 1 which means complete celebrity
51:31
On the other hand if you have 2 or terms that never Walker together because they are completely unrelated than the Idenix would be 0 rather straightfoward way off of the findings document celebrity again there are many many more always to be fined Simoneti by another documents collection of this summer popular away and on its special Époque because it still simple and worked so well for us of a it as a occurred jumped on coalition growth coefficient because it December the idea of statistical relations some way correlation and in the stands things people together so there are some some take the Good differences to mathematical lation but of to avoid the book a slogan summit last look at the example given 3 documents are again he said of set of what to make things a bit more simpler under command 1 step man mankind according to a statement China would be step mankind and we novel would like tool of computer Simoneti by means of the job with it and that means step that instead and that that easy because the documents or in order to end in which you step best the Booker those terms of Protocols at the same time so the simulated use 1 sold on its step and man so we just had the number of documents containing the 1st term that that this case 1 2 3 and then that political containing either 1 of those was all men and instead this contained on documents that are denominated history and now we are looking for documents containing the terms that men contained is contained in 2 0 documents and they are for the chop are index of step and man is to be sold at stood it would for China and men 1st find out the number of documents containing the demand for these China from all those possibly then we have this document and this document because men's contains bells and document new contains rich China no man so he would be to and now we look for the number of man's containing man in China is only 1 documented and so that kind next for the simulated you between men and China would be a hot so left this part of this Matrix because the Dakar index is defined terrific early because it doesn't matter whether I'd exchange you win team this fall below the results of the same which uses suffices to to just computer term celebrities for the upper trying of this matrix and the other of the 2 trying to be a 2nd same and on the day of the said that which or wispy once for these reasons but they now have a method of continuity in the UK and he was it's not to add that related terms to each document are comes complicated found allow somewhat will try to understand how Cape given a document died that is really representative said as a critic with a set of terms than the way the weight assigned to return to the race with back to the document the is 1 minus this product he so that 1st look at some critics to let TBI some term operating in the document
55:44
So this was would mean that when we take the product of a L terms contained in the document that are also we worked as and in some in some factories and their Simoneti between 30 and would be 1 this thing would be the band and the whole point would also be 0 as and that means the weight of which Jerome actually of operating in the document would also be 1 because 1 minus 0 days 1 within the party about their have because we already know that there should be a one in the at the end of the Nineties he said the hallway stunt on the terms of 50 is not not occur in the documents the compel T to are Adblock Duckett to are terms actually the document and for each of these of this time free computer simulating last attempt to that the county to each terms you off the document so this Simoneti is very low for a terms of and the document for some private UK Menachem of about Computer Science and some biological terms like she downright genetics than the Simoneti but genetics and while terms of current the documents would be rather local this means this is a slow and every factors and so this is closer to 1 in every factor and a product of many wants tends to be rather large and 1 mind as a large number tends to be close to 0 in this case so terms that are unrelated to while terms operating in the documents that the local fuzzy weighty so that could be complicated by the bad but on the other hand if it at some time in the document of somebody of science and researchers found when compelling some of this science And research actually focus in the document and that would have high Simoneti team this term would and and this thing would be close to the world and the real life replied with something some other factors which would still be close to 0 and the far the documents to get this behind the signed with high way of everything about this again that are but it looks more complicated than a really is because all this miners 1 and modification makes before event of the but this is the idea tried to try to compel each jumped to all terms going in the document and Dutch how Hall a seminar disputed to the terms of the document and that sometimes highly related to the new terms than assigned to the document the highway politely out but they next example Bolonia is are jicama will Matrix by just use some and he had been out the documents and the previous example became the ones are assigned because the document at the time sacked occurred in the documents that because document 1 man also than men can also document to a step man in China and document 3 staff and main so these are the ones not to the difficult pubs document 1 to what agreed to which I'd do we don't assigned the term China to the document became which take the time China
59:55
It is up to you and then computer products and terms operating in the document so it's the the products its effect as 1 for Step 1 4 men and 1 for mankind the cell use and for each of them the fact 1 minus the and customers that team between China and step in China and mentor mankind and this would be for China and step 0 2 3 and 4 1 minus that is to set for men its 0 5 0 point 5 1 minus 0 1 5 half and from mankind 0 and 1 9 0 McMonagle yet to 3rd FIA China gets signed with the high which was document because of made many because China Calker that with step and many other documents Chabon man seems to be highly related to to China we know this from document to and because we also have stepped and men and women want be a sign China to 8 with a degree of 2 of of works much pedological lectures on really really have some some that would be very quickly and allies to find semantic relationship between between times that the General idea just assign the words which documents that are related to the times already contained in the document so now we have to wait and a way to use bullion boy in the family fabulous to to ask in some way to tegulated results by using the new in maximum an hour complete the to model that we can use
1:02:23
So what are the disadvantages spent advantages of this found So as seemed the computational these the membership wide weights usually on tends to be quite difficult because it free if you 1st to compute the simulated between any terms in each of the past a new collections and you have a number of firms times number of for a number of times computations this who is usually large for off last recommend collections and assigning these rates of automatic because when you the father other tree overweight must be within 0 and 1 so we couldn't just simply use the number of times occurs in a document for doing this economic rise in some way to the ruins and the and 1 needs to have some some clever methods to do this but it's not too easy actually so that have seen on the might be problems with intuitiveness went discount comes to clear possessing granted Osnos but this really sounds will never by the cell scenarios usually on some in some kind in the spirit of databases information and feel the need something that made me want you to stay in the state of NEMO and Maxine maximum so that is some found some advocacy of the said using the so called Tinos anticodon almost to define the identity of the name of the maximum that has dissatisfies amendment conditions in this but not Caunos can use some way to wait to compute discussed but usually that sets not unmanned and doing fuzzy logic of not to popular and information retrieval of dozens is used in this way 1 of the advantages of into lot of cars will not be able to L use non binary term assignment documents which is very intuitive because some terms are highly difficult for document a mistake and some just and And this could be affected in this way it's only and but doing this as a team we are able to find documented are related to create terms but do not contained some of the resurgence Isaacs on the and so big advantage is being now Frank results that we just can't computed not just of the results and put those documents on the talk that have a high grade memberships called when evacuating are created from basic and the 1st step for 2 with 2 walks the search engine as we know it
1:05:22
Right from the US and some that are not for us to do this 1 of the 2 other firms its its is about their philosophy behind houses launch so that these thing to say about just 2 weeks ago about the by remember ships but stadia grave in some ways to win could be membership agreed in some set off 0 2 point to 5 but what does this mean intuitively preceding the and began to some categories regions with all this only things but what the meaning of this sort that possible them a many husband applications for some it could be some ability sort it if they some think something X is contained in this said it was degree of quieter than this could be the this contained with the ability of a quote so in some cases contains some cases as much as cooperate is seeded than it could expressed missing Norwich while climbed to 25 per cent sure that the banks is contained in the centenary year might be one way well on the way up just a question of whether it is contained in a small part of ex could invitation million something outside just 4 but that was as complete nonsense and said this is what 1 most people thought when the but Saudi proposed is ideal for the magic of the distance from the Fifties and Sixties the thought of this man's crazy for doing need grated membership of the don't make any sense found and Saudi also all worked on a on a intuitive interpretation of his idea so and and his 2 colleagues said describe so called possibilities so for the to a statement to the problem as 29 year sold Soldinger young and then possibilities about the degree of compatibility so is 29 young doesn't sound is reasonable and the degree of compatibility is exactly what fuzzy membership agrees try to measure told reasonable the culmination of things all so the signing of the summer research to a document about science with a degree of 64 and silence reasonable to 64 and Mary so that the British that usually used in went using dealing with became clear focus is on the precise concepts because you can not defined young and it was this bad young his the deputy debate concept is it that so far the that the causes of the father logic because you cannot you can give your boss of each line serious 100 year than youngest long somewhere out here to you may be 1 starting here but it's not easy to define board of see so it's about imprecision and take them by the logic is not about missing knowlege so when wind assigning a degree of comfort compatibility of 29 to younger than don't want to express all I'd don't know or youngest defined assuming but there are some definition I'd just don't know what it its symbolism is saying that there was no clear definition so it's not that don't notably the definition that there is no definition of this squad natural to Qumu language because we are using the statement all the time that this is rather large he's quite big of something in the style of play on the 1st one off for rich examples used a one off his papers filed and it takes X for the so and he reckons he the difference between possibilities employability that we have some background knowledge about hundreds and we we know that he usually he to X sometimes he had just won a for breakfast and and on which 3 but he never its more than 3 stone affects No 5 6 and not even more so in the knowledge that can we can constructive function that yet when shows some random letting the probability of a depressed and he will need to X because only to act and that the Heat's the probability of campus and only 1 and a bit of campus and exactly 3 acts that probability possibilities would sign these numbers its computer reasonable to assume that he may be wanted him also made 2 words the beatings Rix's all possible for human beings FedEx yet well on wheels with my victory where X Men things become come complicated and then you try to read the small expand you can imagine what happened and some sort can also see that the new possibilities you don't have to rule to some to some the road to the values don't need to some up to 1 in this case probability as you know all this need to some of the 1 and two year round meeting with the completely possible that no doubt about this month might have to work but that someone each other at the time as it did excesses mounting them Europe that I would be Republic and so as to possibilities about Sundance possibility of probably or combine in a way that possibility provide some some after bold for publicity the probability that this is what actually happened and possibilities as well that completely reasonable that this is a function without a side in the wake computer point each year we could easily see because that's not very possible that this would be a very clever Simon members of the that's all possibilities used sewn up some good the summer this time thinking about a glass of water and a assuming someone just gives me a glass of
1:12:23
From clear and I don't know what's in it but I'm not its use players what it's what they don't know which it is so improbabilities period he would say Well the glass is of players of World War Two talks on Blue my degree of the these that this classes for of poisonous 20 per cent so that actually might be the case but that but it's highly likely that it is only water so my chance of dying he is only 20 per cent so that even if you do decide to Gaza it is and it is is either completely for what completely full of present impossibilities the read he could think of this as a mixture of both so it could be but there a bid water so it could be worth 3 poison appointed water of this possibility possibilities he re tries to twice before midnight in some way a care will next declined matching is a quiet subtopic and and make a small break a cable and but just that will set of words could so each document is just about just about the only country in a by binary refashioned but revising the six under the this which by complicated to use and paternity are back of words really kind of queries used by search and Lifebuoy to just type in a set of key words you are looking for and don't want to use of some some complicated family and than on just documents so whatever you put into the search box of his search engine in some way is the the document which uses a very small document that it is document and that it is clear the idea behind or if you want some time preventing the following and this approach goes back to some idea of Hunt's but and 1 of the also on the last lecture Flybe and meant to be 1 of the 1st the measure to a systems and his Nbody was when someone is using the is the before document and this person should do is simply should write somewhere that describe this document and then we compelled this description to are documented the database so queries about command and that the doctors and that a document and now the only need a method to compare and documents possibility but don't need to distinguish between crews and documents and don't need to do something less some special a handling of the ways but just can say which created documented now we only to compute celebrity between documents which could be made quite easy and this time the December Famba and message was that are based on the type kind of Greece and as you know is the son of free rein in all and and so caught 11 matching is that it's a very simple way to want to answer the Beckham spurious and this and this and this
1:15:53
Some document in the in the in the in the in the collection has exactly and different in common with to create the the relevance of this document is or can can be measured by the number and this document has only and minus 1 terms in common with you create than the 11th called the and minus 1 display sought and the coronation that is also cut displays of over here and you just don't a number of firms that through and a document income of actually really easy and this week would be on Sept just by citing the document collection by this 20 communication level and and returning the 20 documents having the High School of Music hitting the largest below that with your Creery document again and some of the documents the grid is man and MannKind and now we are looking at the that between which document and a group and we see that in document 1 man and mankind's contained close to lower its 2 he on menace contained or this 1 is only mankind meant it contains over this 1 and they are a ranking would be document 1 in the 1st place for scoffed to and documents to win 3 on the 2nd place both was called 1 never criteria Chinaman mankind has already of Turin documents 1 2 and only of 1 in documents 3 annoyed at Franklin's really easy that most simple of way to do this kind of Greece of found the uses densify and are the most potent and the most popular that popular a given more Lazaridis displays more which will discuss after the break out in 5 minutes Orion that led to continue with the victors space smaller so I'm this probably the most important will you on a need to know intimated fewer and actually if you Beijing and made into the system on all your own this published the 1st approached your you're going to try again and it's 1 of the earliest and most successful approaches also over the idea of allowing the practise based model so that comes from from the the information spaces for this for somebody go to library and usually the books that are related in some way or standing sidebyside so he this about computer science or there as a child about genetics and on the other end of the libraries decision about psychology and document on by topic in some way and and to the and the idea after Bismarck hot whether we could transfer this location principled to integration with model since always so just Group said the documents to together in some of the some space between them and may be much more adamant that some of the documents about the same topic should have the same locations in the abstract semantic space so and instability between documents could be measured by some proximity measure that sucking distance effects sampled the distance this man and this distance of Priestley's large and as we are now using back of words could you each Creeley's a document and we just can't compete release to document as we can but that documents to created to documents really think we just need a clever presentation of all documents in some space to some coordinate look at it as the most players but also the and the famous certain to receive the award named by himself to 1 of the fathers of automated fuel and he that is the documents increased public presented as a point in time and that means the that there is so and usually very large like half a million or so and it is the size of its vocabulary so that is each time that are could somewhere in New document collection produces an access in this space
1:20:42
So if you have 10 different times in your collections Dakota's more collections and tendrement space and page document is located somewhere in the stand the main no space so and hobbled at terms access yet wellies Symbian used to the a number of occurrences beachcomb has gone itself up some good if the term science because 2 times in document 1 than document 1 guest at the coordinate to on the excess assigned to the terms of the so crowded of the incidents that dose of documents so of to back some of the document 1 contains the stock timestep in China where that China occurs 3 times and document to contains also step in China and step across times in China only once and document 3 just contains the terms that only wants and then 2 were found dead in China not collection with grant access for of a restricted to ecstasy because of painting would be a bit difficult to with more access and then because is the appeal of the documents in the space of a document 1 would have to coordinate 1 on the step excess and quoted and read on the China eggs and the same could be done with documents and documents that are related in some way or close 3 2 3 2 idea so you can easily see this in the next because so small but documents and that the very same about tend to have the same quality space and also tend to have smaller distance Simenon somewhere like a non to define not family defined simulate your approximately in some space of some some tools from mathematics from which is called the metric so a metric on some said that is a function that company to elements of the said to each other and returned to the ring number that which she should measured the distance between the 2 items so that some properties method that metric should have won this non negativity these numbers are 0 positive for the next 1 is that not team this values should be exactly that
1:23:27
It was the immense to be compelled all the same so that compelling combating document 1 to document 1 than the distance should be of calls 0 which of men because the distance between document London document to the same as the distance of the document to and document 1 and it should satisfy the the so called fighting in the quality of effect to points that want to compel then this the of baby should be smaller now than if my office point during some some way around here would have a distance between a and and distance between and the and the and the means that during the direct way is shorter than going and the 2 of so them that the finding the quality some basic mathematical features them man method of distant should have and 1 popular some of the human distance just take but it is recording of those are the demands to be compelled to take the difference at summit up and take scrambled of it and that's exactly the kind of difference reusing federally in space so it you want but a what they do at meaning of sitting in the stands
1:25:02
Palmer guide it simply it simply distance and space as we know it it is a sad talks sound some where documents on this site at a distance of 1 from document 1 0 4 was a boat circles and that would documents having the same distance from so I know this from school of quite easy but another concept of this relocated to Metrix is simonetta came as a soul matrixed measured distant the logic of the more December last some and Simoneti measure measures Simoneti again we compared to objects and now we have a squad that ties between 0 and 1 0 among the means that these 2 objects are Maximilien's but some but they identical and 0 means that their maximally December now so far completely on related to of it all was between 0 and 1 and not just a real number through the metric OK and that this summer is not the only agreed to mathematical read about whether the simulated measures should have regarding probity found out that the US wants to put the measure of the cost a cosine Simoneti electors basis and it's simply the Anglo between on the quality of these 2 points for the thing that makes some players that yes so found in a case found 1 compelling to documents for example document 1 and document to adjust line from each documented the origin measure of the area and take the car to sign off his anger that the anger is between now 0 and up 9 degree and since Ahmed should American to measures should livened weaned 0 and 1 way or 1 means Maximilian Scimeca with Texaco signed and the car signs of 0 is 1 so and document being very sympathetic to the document 1 would have a very small part of a new year and large coastline documents which are highly December Dutchman a year and document the human would have anger of 92 Greece and the Coast on Monday Greece 0 the Stockman's a Maximilian when or documents lying on this line but equally cymbalines similar to document 1 for some of each other document that that's a general idea of coastline Szymanowski a chaotic computing the between 2 actors and brought drawing it and measured it seems rally the follow with so we need some mathematics so we be know because signed off the spangled his the Duff contact between the 2 Lecter's so of the continent of document a and to the extent of the document wine taking the product that simply some or all poets taking on integer coronets
1:28:47
1 Germany it's small beyond the name scale product and then divided by the length of both that just because the legs of a doesn't matter as long as the direct is the same so sounds pretty obvious from the length its utility NAMD just take this each coordinate summit appendectomies Greta would be easy on the amendment of racial so did the determining the similarity of the coastline Simoneti between 2 vector was simply means Computing this determining the length of the beach backed off and the fighting is thought by the length rather he using became Luke now can come back truck matching some because there is only a special case of of that just displays Möllemann using collapsibility that system for the moment at this light only that are term but only concerned by which time occurrences so vulnerable where it and then discovering a product of the new Vectra X and the document of actual wide is expected recording nation Levitt of X and wide because if these things only 0 1 or 0 0 then the sound is simply and the size of the old that between the query and documents the number of terms of Corringle so quiet matching special case of that the state model with cold and celebrity working hours them cinematic specimen levity to measures that pleading distance for some of that is because I'm a Celebrity them many many many more Simoneti measures so which 1 to use all usually this about a typical think information feel there is no correct answer to this question but is depends on the type of dedicated dealing with some document collections European distance could be a really great ideas for some collections of the celebrity could be a great idea since but find somewhere in the future and usually you have to try what you want used so and the most visible signs of a mostly is using the and goes on similarity because you can Nickleodeon distance has some problems and usually it is the case that to do different measures could you could use the Hayes somewhat simmered but not waste of books of the year it is a comparison between you can distance and and celebrity and as a concede a found if the if you have some of the 2 to documents having quite depleting distance so between these 2 point the distances rather than or so the covert will be rather high because the banque is more of this case but it might have been that if you have a highly cosine Simoneti then I'm the Keegan distance could be completely different because if you have some somewhere found than in the stem engine that could be last opinion distance although the anger about his rather rose more of these measures usually out quite a kind different and depends on the type of of document collected dealing with
1:32:40
A machine that could sensibility does not depend on the length of the documents and and Greece you dealing with so it doesn't matter whether documents the document focus step in China on each only once the China and step across tries of the document or 3 times that it was always the same victories and for measuring coastline celebrities it all with the same sort of instability focuses on comparing the relatives times amounts of how often each time Walker's of its and it is about what we know that Chinalco's more often than step almost Walker the equally after the document but it doesn't matter how often data the of sound also referred to the length of documents so if document to his just document 1 plus a copy of document one paced at the end of the 1 and after a while he won again and this is called the to spend it with the same properties regarding cubicles time simulating so but using you can understand this to documents would be different because they different length so and depending on No 1 measures you use it might be important to distinguish between documents that different length or not from you could also use some not realising function to account for document length most popular his focus on the dividing each corner by the by that of length and then they should not be a lengthy 1 that so the again on example
1:34:40
2 documents which are But couple's of each other and it will be as well as 2 legs 1 and is 1 would may be the man to this point and this 1 will be made at this point because the dividing each coordinate bloodstained number and then you get a links 1 victim both bisected Endeca and and in the European distance which we both documents as being the same as as a divide the granted by the victim's largest coordinate or by the side of squad and so when using the nomination by the length to use this grand he is where he and my squirrelled so that the different difference between these 2 sides met of taste this no clear rule out to it follows usually are 1 doesn't do any no money nation and just applied the coal sensibility because it independent of any length issues if using measures like you can understand the need to care about no misation Angels and and a collection of a care special case is far 1st option rising to you and that the US and in this case where documents and furious after no moonlighting are located on and on a walk around the world and Adam engines on a on a unit of the unit said it still high this week about the the around the origin
1:36:32
And in this case You can distance and signed simulator are not identical because if if you compared these 2 documents they have simulator T that depends on the state of and if you measure the coastline and the European distance between these 2 of these 2 it also disproportionate to the full sentiment so not rising to its length volume of no more produces a document representation in which you can distance and go sensibility identical so tricky found became and sometimes not is Asian is not a very good idea because if you have not documents that might be the case and that might indicate that the cover up that some topic of morning death and at the other end of the book about condoms Physics this problem Madden led by the more helpful than just 1 page summary of the idea that you really want to know about and Physics about those documents could have various a presentation that space and only days from basic early in length so of that the 2 would be to account for off a document length in some way and Soha this be done about 1 1 way would be to compute decreed resolved on all novelised documented the and the and give state documents some somebody else in the ranking that they appear higher in the file a ranking of the pasta to the length the costs and the her on documents at the the beginning of this and the fact once at the end and if you have a highly 11 Chepstow command and and not sort of and undocumented bulls might might appear the same at the same rate position in the results from what after could do some you could measured the effect of document length on the rather than and that current collection and then try to determine last actually the picture effect of document length on Naunton informativeness about these documents and then menu after did new some some clever to do this you could all used this morning to with to be ranked you results are quoted the length of the band's owned what you going to do with the collections and how the collection at the tree looks some of it but it does acknowledging because only the religious frequency between document terms and the matter is that it seemed that sample of the many documents that only different length but not in appearance of terms than they would have to sample sensibility to of this money and it is only a it didn't depend on a single termitic hands on all along the frequencies of different times of day to day of still if it with a documentary where I want to because 10 times and the other turn them the 1st 20 times than fantastical sensibility it's the same as a document in which the 1st term Walker's only once and that some of the coastline of the length of the land is owned ignored apples and simulating if the continent are the smallness of each other but since mid even with the back of with small taken that all of the documents that look like a bit of each other could make a big difference so that could be last documented the news will be much information it that was essentially a in the back of words smaller just the 1st document which tries in a match because the different from a different point and so that the public are paying a high Walker Cup could could be used on all of this is back to representation full ranking for results and and but and also make the following observation some words of repeat a document Booker very often then they seem to be compulsion of calls a month Physics output the document about history that probably the term history would occur February off and found in the back of what followed this this is expressed by the so called time frequency of the shot and the use and patient yet of the and this means that found the time to book a as so often in document died just frequency and but also have been recognised is that it would have a large collection of documents than some words might be highly specific and some words might be might occur in all and in almost every document for example a few words like the all N off I Introduction and scientific papers that terms that popcorn every document and and are not very typical for each document them so he and the some police have crap assaults like of offers of documents about psychology and the terms psychology might not be very directory sticks to a document if it some documents contained the Dome psychology but it of his science Crepis if there is 1 document containing the term psychology and this document seems to be special and seems to be a rather than of psychology because of Psychologies the rest psychology
1:43:19
A couple of 1st much more often than in a typical document of this collection and so that ideas to measure the discriminated part of these terms found 5 tries by some number of some discriminated part of psychology would be in a psychological of call and high and the design of found that some were stood ephemerides it but in general I would have you would like to have the following if a firm would like to have a child at a time frequency and with the term his and has a high discriminated part is more specific Benhua would also would like to have a high term weight in general so that they could ideas to which terms by the the time frequency might declined to fight discriminated so the psychology of workers in the UK in the designs document and it might be equally that we could be corrected stick for this document than the term computer which also could be a very often in the collection and since computer focused and almost every Duckham and it's not very very difficult for the collection of a this principle was full of life by Kevin spy Jones and Seventies will introduce that the idea that measure of the idea was that this specificity of that is exactly what the discriminated part was about sort found the question that has been home any documents some is content and specificity system was negatively correlated with his number solicitor more cursory very rarely a collection and it seems to be very specific and book it occurs everywhere and it's not the specific so and left the city determined not to be discriminated part that it can be used to distinguish different documents for the tactics of documents said the psychology and in the computer science documents strongly indicates that this document is related some of psychology so I'm to this led to the development of the team idea made up to deal with the document frequency for each time that's the number of documents containing a given time to see this measures the history is is physically of system so and the back Jones a proposed the so called Tea idea after waiting scheme until the weight of each time in some document died was simply the Tempe could the document so how often this term walkers and the document the violated by the document by the document frequency of system and divided it's called the universe document frequency and there for you if you idea so sometimes get a high high highway for document if it occurs very often that if UK is that the reality of the documents because of the coast and in the document that it seems to be very hot because other people are not talking about it and you could also use a a more refined waiting scheme on the spot Jones proposed that the relationship been between specificity and invested mindfucking to should be logarithmic followed if you have a Ound the the that will amount of document frequency that should only called as a slight increase in interest in perceived FFT specificity and this something I'd something from the rating scheme that just works well so would also adult the use and the other 1 not follow ejusdem momenta of smell of designing the order of the day depends on a collection you just have to try out the usually found that the most common form of the idea is this month to take the time to do the job preprint how often atomic in in the document and multiplied by logarithm off the number of documents in the collection with was the smaller it added weight to the calls of every 3 career at times so that these logarithm never get to the world for this autumn and never get 0 0 and the document frequency also plus 0 1 5 2 2 2 to get results that that could could today document and is so the and divided by the document frequency not as a suspect collection size of this because it is here that the amount of documents in breach of the terms of the that everywhere that's not worth mentioning if you time that are very real used in the collections nations that appeal and this is a big document and then this time seems to be part of the way it is used comes from looking at the time I'm getting this fund is not approach to to time discrimination approach of budget to analyze disability across documents and try to find out what the what the influence of each individual time when Simoneti and try to try to measure the terms Houghton's by the effect of the real used to want to be but he moved to here and now we have seen some some different models poetry that could be done by the end of the and that just days and of the of of the Eighties on Joe softened and his students and highest hybrids the results of these different motives are high this number is how these numbers are computed will be topic of 1 of the next lectures on whether a pledge is that higher numbers of the up and means that a result quality of on Sunday and document collection for of medical documents and these communications of case I am some science Java a defiant some example curious and and compared the results of the IRA the 2 at the humans who would rather than and if the results are very close to what would humans were just rather and then the Cisco idea would be rather like so we can see that the practise was outlawed usually get the best results you with some some large distance to the other with Parallel and the folly retriever usually are not that good on these on these 2 piccolo tax collections found and so you can see a time when the display smaller was was invented it was it huge step followed because of quality freezes increased lost a did not understand the advantages of the but just this log on big advantage but it's easy to understand document appointed a point in phase and cooking is just by writing about command and comparing documents in the collections of the and as it was a scene it can be highly customised to the new mood collection for example you can use different distances Simoneti functions to member to measure simonetta document or wholesome criticism document you could apply nominees schemes for document length you could use different methods term Whiting William cupolas not some some other awaits so that this is not basically provides the framework about it but you can plug in different different functions and different schemes and different term writing methods and you guys have to highly human you to rhythm to the specific properties of the collection so far this year is rather limited but you can do with it the only chance you have is changing the way weights are assigned to the but few volume exactly 0 degrees of freedom just takes a times at the appeal the documents and then crude would fall that there is no way to change anything in the UK as a small your highly highly flexible and what's left most important information feel it's just works so of results a pretty good and bad because said that after most of the used model to the arrangement fees and most probably fewer if you if you find some some days I would be system somewhere and they will use paceman in some way so was also pro said the that is possible on that display smaller that means that when given a list of results users can simply say iLike this document and this document and this document is not a man impress on his feet that you can't easy compute in New list of results with make displays what it was about what is advantages of course you have to deal with quite high dimension of that does basis 0 each Tom is and excesses and each demand has a lot of clout in and you have to work with them in some way to go yet but the computer trucks and for this you need some specialized tourism and reduced in the next lecture of the lecture after that
1:54:01
That displays on also relies on some in substance which usually on Monday to find out the stated that some the of other this which should be means that are documents that are have a small possibilities are located in similar provisions in space tend to be rather than to the suspect to the same pre of this there is a case about the idea that we can begin applied deck of what's wrong for presenting documents so of Qwest it could be the case that you have to documents with the same lack of Russia presentation that has a completely different meaning because at the right place is there has not in the document and his completed the although opposite topic of the opinion that still these to documents have the same representation in 0 space is a cluster of another assumption is the independence such assumption and as we take 1 exes Patel and X is part of what to each other which is you that the equivalent of 1 term is independent of the current of another so this problematic when dealing with the news of and and because of the was in the news those terms all related so if 1 terms Corus than the other 2 also tends to be same document so these to excess should be 2 independent excess and there should be just 1 excess for these synonym to comes out and 2 weeks we will see a way how we can deal with that away came not so the talked about Manuel's automatic indexing so Classic in a new library signed a Runtime that evidence the only that the next thing of documents that means assigning curious to doctors to documents and was to do this manually like is done mesh investigation that some very switched to be assigned Hewitt's to documents that seemed to be highly related to to the documents and the best way to do it is to do it and I met my 1st time in my not every fish and that the idea was that the result quality of the quality of document description is really really high that Manuel work it's worth the effort soul and modern eye on Web search whom you an automatic assigning Nextel's documents just by assuming that each word in the document as the next so the all back away at Loughborough which document is that words and interested libraries you would just if a signed for 5 most of and if he would like this to the kitchen systems and he are efficiency so doing automatically doing the cost is more important than finding exactly the white Hewitt's to describe the documents and the something strange happens that you that you buy some reason to choose the wrong I next terms because some strange occur the document that are physically under which is a topic that this just happened so that we can get used to the cost of basic that these 2 lines of French doing good document description with a lot of work and doing cost document ascriptions with that flows quality of description and in the 19th 62 situations who was last like this you could you that she was a high quality of fuel of after that index with a lot of law of only could use not emetic indexing approach was assuming low quality that could be created easy but you really want is something like this and the creative both high quality next zone and the reserve Preston at this time last highly could about being out of the findings menu indexing processes to get at this sweetspot where I really want to want to be so they tried some some clever classification methods and tried to buy the paper said the document that you don't need so many experts for somebody could you could take a bunch of students reclassifying the document and and the expert only do some fine during which could speed things out of this 1 line of research has done at the time in the Sixties on the verge of talking was taken and the and the Cranfield research project and they are now want to look at it expected this problem how to make the indexing more more effective and efficient at same time so on data investigated 29 of in mixing and which as a child in a world that seems to index documents Badda and cost so and the suprising and and and and they are also evaluated some information to the system
1:59:33
Regarding the quality of indexing downhill and the review apprising without loss that all the automatic indexing links to the east is good results scale from annually mixing so that so it Computing you don't have any advantages when trying to rudimentary indexing because automatic indexing already used up their not stonier so many indexing was in February to automatically mixing it noted that the 2 companies this time so the workload and was the leading scientists in this granted project and some such occasions for which he are made some quotations to hear some use after discovering on this results and books on with this computer is so controversial and so unexpected that is bound to stroke of on the message methods which has been used so I couldn't do it I couldn't believe prime time so that has automatic mesh could pick up a bat and a menu indexing
2:00:42
So they decided to do it a completely shake off the automatic annexing approach that everything seemed to be correct so and since they don't didn't find any any possible problems he found there and they had had accepted the result was rather surprising because and as said it does not the crest except to look at time to explain the results which seemed to offend against every can on the between faintest librarians of says that the audience has dedicated strained that early manually saying that the best way to do it and he becomes the project and it proves that the automatic indexing it's just the way to do it so that the audience become it is fully mixing him another approach was of this not system also opened smile stands for system for the mechanical of wheat into trees the of text and there also also last information to the system to the Sixties all again 1 of the 1st approaches to do well processes documents automatic early and make is fallout from collide Reese to automatic possessing of documents that the tree was bound to get out and funds are not the German immigrant into the last and some people's say J sold was information extremist 1 of the major figures in the field for off for several decades and not has 1st has that implementation of that there is more of an possibility that produced the and some weeks from now
2:02:39
Follow his some some pictures told about how well the use of computer was just became quite popular in large companies to occur to them calculation and is an Ibn 7 0 9 4 which actually used for this not project and usually though is that this year some comfort des with some with some die from my Ibn and in the suit and it gave him Royal programs and he I put somewhere is machine and this or Computing worked at this time of year some citation about the speed of his machine and the machine a bracing for breaking cycle that was to Microsoft to microseconds that are that out 500 operations 2nd and 3rd and all at the 5 million right and the 500 thousand right media thousand micro is 1 of along so this actually is 500 thousand 2nd this is 500 case was so if you compatible and stand up locally of the acts of a rather slow machine but this has not looks like the text in the face and was developed under the mid Nineties and he the of summit Sunbirds but you can read it in the document collection of the new can Kanchelskis's statistics on not under the structures created and then and created the index containing containing all these in the fight and then you can you can do with to run in this file you had to specify you create and then you to you a couple results from not systems and of Qwest as set when no would be with information to create a few are while with need to evaluate Huey's on quality empirical the usually 1 1 uses some tests collections that publicly available and it is not systems are Latchley at the last variety of this collections used for example to a but popular John of designs with most surveyor on tickets from from many different feel some collection of from Ivory signs of the times magazines are in the rest of the board range of different document collection use Tessas not system and the way the system is to be the point he has been has been very convincing so that act as a small it doesn't just work on the city collection but it does work for a fine earlier this summer because of the ability to get a prepared for the next lecturer because discuss publicity in the next week and will have a look at some some concept the need for this soulless ability independence and is appropriate in a Syrian indexes rather quick on so you can look at up told again to be the but next week and the probability that the Phillies sees the likelihood or chances that something will happen as it was seeing the fat 6 sound and usually you have somebody fine random experiments for example realistic cited by his than road again and if you read it he's 19 where the physical rose 1 than you win all that sort of game William my played and other white use and publicity which is totally has been has been devised to compute the winning chances in games like this 1 in 4 of the 6 other game found and you can rise this game by looking at all different events that could have happened this the 1st words of the ties is the 2nd row selected hasn't happened can 6 times 6 6 different events and you win if you have had used 9 interocular while you win a few second row would 1 jaws we and you will probability is just the number of winning event divided by the number of quoted and so you have 36 events Interbrew and 1 2 3 4 5 6 7 8 9 10 and 11 30 13 14 15 16 million and the winning of the beauty is this year and is less than half of this ability and so that the was a the downfall for some of the major events of some of the probability of these 9
2:08:19
This area these are 10 different events divided by 36 the probability of a ball 28 per cent probability of getting a 1 in the 2nd round is just this Colomiers probability 17 per cent and viability of winning as you've seen its 25 per cent in total told The Independent that it is a concept that would be to leave and that might happen and 2 events on the independent is that the currency of 1 year and 2 dozen changed the publicity of the other went so for some good the study finishing to events and the independent if and only if the probability that the book is exactly the probability that 1 occurs when he got back from the 2nd Monaco's found and he some example and are these 2 events independent 0 in the 1st round for the 2nd row of quiet they are independent because the to simply does depend on the point of the sale and other question is whether to interview and 6 in the 2nd row where these are independent events and these are not independent events because it's by far the number last 10 interflow than my 2nd must be paid used for off so I'd already ruled out the situations 1 3 2 and 1 2 and 3 for a 2nd round of these are not independent and the same is true for the on the trade in total metals 5 the 1st round because with operating on the left interview it could never have been happy that had 5 in the 1st robocalls getting trials means 6 in the 1st row and 6 in the 2nd row the other way around defender had 5 in the 1st round by can be sure that I'd don't have tried in total of these events are not you can sue the probability of this exactly the probability that the and is a big event occurs given that already know that some of the event already or so for example that way the probability of winning the game given eye for in the 1st where so that not only talk that look at the event with the fat and the 1st and only 6 events are part part and here and there is a winning events and the winning events and so the probability of winning the game given eye but from the 1st round of or I know that is not so of probability of having had a flat in the 1st were given out when the game is just the other way around so given Iwunda the game is already know that at 1 point
2:11:32
Then on somewhere in this area he or and there in the 1st round would be only the 3 events of the probability of having it for the 1st goal went and that when the game is to be divided by the number of events he said 3 voted by 16 this 19 per cent so that condition appropriately mathematical terms the probability that there was events occurred divided by the ability the event and the and the event with a condition occurs and find a way of pace really and some time after a Tomas based lift in the 18th century and he found out that the probability of a given the is the same as the ability of a Land by the ability of the times of the ability of the given a a busy a found a way to relocate these 2 we went by swapping around the condition and the event sulphoxide the the probability of having it there in the 1st round the given away the game
2:12:50
Has over the area the ball in 1st given 1 and that contrary to the family of this should be the probability of 4 in the 1st divided by the ability of winning timestep all albeit of 1 divided by for the 1st And this 3 of quality care Veronike's on A half divided by the time the and this makes from 19 per cent in the end and has already had these example of what was and are correct locate found a is usually called the profitability of the game because of that the the probability that a or crows before of new anything about what happened and the squad supposed to the ability of a request that the probability that a or close after I'd got some new information and the commission was that the event be book a so of the price is minority before have some information about the the general context and the and the poster problems probability that they Corus after Miller something about the situation so that when I'd get some new information while serving the than the probability of a gift updated to be and it appropriate figure in in the world this idea of pride was Tarriaro but that it next lecture we will take a look at for is a good few more world and are on the banned Ndiritu must be thank you
00:00
Area
Polygon mesh
State of matter
Electronic program guide
Feedback
Physical law
Scientific modelling
Computer
Mereology
Cartesian coordinate system
Computational intelligence
Area
Web 2.0
Information retrieval
Computer animation
Personal digital assistant
Website
Physical system
Physical system
World Wide Web Consortium
02:01
Complex (psychology)
Structural load
File format
Home page
Complete metric space
Computational intelligence
Mereology
Area
Computer
Video game
Computer network
Electronic meeting system
Formal grammar
Physical system
Area
Covering space
Computer font
PowerPoint
Numbering scheme
View (database)
Firewall (computing)
Computer file
Web page
Interior (topology)
Computer
Bookmark (World Wide Web)
Open set
Information systems
Simulation
Physical system
Fundamental theorem of algebra
Resultant
Point (geometry)
Domain name
Polygon mesh
Numbering scheme
Proxy server
Streaming media
Hand fan
Firefox <Programm>
Landing page
Internetworking
Term (mathematics)
Database
Maize
World Wide Web Consortium
Window
Raw image format
Electronic data interchange
Polygon mesh
Slide rule
Online help
Server (computing)
Cellular automaton
Electronic program guide
Magnetooptical drive
Expert system
Coma Berenices
Euler angles
Grass (card game)
Shape (magazine)
Error message
Computer animation
Personal digital assistant
Information retrieval
Synchronization
Address space
03:46
Software engineering
Proxy server
Structural load
Home page
File format
MIDI
Area
Landing page
Computer network
Lipschitz continuity
Information
World Wide Web Consortium
Mobile app
Metropolitan area network
PowerPoint
Slide rule
View (database)
Firewall (computing)
Server (computing)
Web page
Scientific modelling
Computer
Information technology consulting
Coma Berenices
Bookmark (World Wide Web)
Sign (mathematics)
Error message
Computer animation
Compilation album
Solitary confinement
Units of measurement
Physical system
Address space
Wide area network
05:27
Multiplication sign
Scientific modelling
Mereology
Perspective (visual)
Tracing (software)
Web 2.0
Information retrieval
Data model
Word
Mathematics
Sign (mathematics)
Video game
Roundness (object)
Matrix (mathematics)
Moving average
Information
Office suite
Physical system
Area
Programming paradigm
Logarithm
Algorithm
Theory of relativity
Constraint (mathematics)
Numbering scheme
Sound effect
Hecke operator
Bit
Term (mathematics)
Arithmetic mean
Data model
Network topology
Database
Order (biology)
Information systems
Right angle
Quicksort
Mathematician
Task (computing)
Resultant
Point (geometry)
Purchasing
Domain name
Readonly memory
Polygon mesh
Server (computing)
Game controller
Presentation of a group
Event horizon
Number
Latent heat
Causality
Term (mathematics)
Natural number
Database
Energy level
Subtraction
Units of measurement
Condition number
World Wide Web Consortium
Data type
Information
Key (cryptography)
Expert system
Set (mathematics)
Library catalog
Table (information)
Word
Computer animation
Personal digital assistant
Search engine (computing)
Dependent and independent variables
Object (grammar)
20:08
Readonly memory
Structural load
Branch (computer science)
Mereology
Matrix (mathematics)
Field (computer science)
Escape character
Internetworking
Term (mathematics)
Data compression
Database
Moving average
Distributed computing
Representation (politics)
Data structure
Condition number
Physical system
World Wide Web Consortium
Area
Algorithm
Information
Electronic mailing list
Term (mathematics)
System call
Estimator
Stochastic differential equation
Computer animation
Network topology
Search engine (computing)
Smart card
MiniDisc
Key (cryptography)
25:22
Axiom of choice
Database transaction
Query language
State of matter
Multiplication sign
Price index
Weight
Mereology
Matrix (mathematics)
Word
Information retrieval
Sign (mathematics)
Video game
Positional notation
Bit rate
Hypermedia
Damping
Position operator
Physical system
Social class
Area
Metropolitan area network
Binary code
Basis (linear algebra)
Electronic mailing list
Sound effect
Bit
Term (mathematics)
Bulletin board system
Degree (graph theory)
Wellformed formula
Network topology
Order (biology)
Information systems
Quicksort
Resultant
Point (geometry)
Slide rule
Set (mathematics)
MIDI
Online help
Event horizon
Number
Frequency
Term (mathematics)
Operator (mathematics)
Database
Lie group
Subtraction
Condition number
World Wide Web Consortium
Information
Pseudonymization
Consistency
Element (mathematics)
Content (media)
Set (mathematics)
Cartesian coordinate system
Local Group
Compiler
Subject indexing
Summation
Computer animation
Logic
Personal digital assistant
27:36
Metropolitan area network
Query language
Sine
Structural load
Set (mathematics)
MIDI
Price index
Binary code
Weight
Term (mathematics)
Matrix (mathematics)
Maxima and minima
Word
Information retrieval
Escape character
Wellformed formula
Computer animation
Fuzzy logic
Element (mathematics)
Key (cryptography)
Fuzzy logic
Simulation
World Wide Web Consortium
30:04
Spacetime
Set (mathematics)
Interior (topology)
Scientific modelling
Theory
Price index
Binary code
Weight
Matching (graph theory)
Term (mathematics)
Icosahedron
Information retrieval
Data model
Computer animation
Vector space
Fuzzy logic
Element (mathematics)
Moving average
Energy level
Fuzzy logic
Uniform space
World Wide Web Consortium
33:10
Asynchronous Transfer Mode
Spacetime
Set (mathematics)
Scientific modelling
Theory
Price index
Binary code
Weight
Matching (graph theory)
Term (mathematics)
Open set
Information retrieval
Data model
Degree (graph theory)
Computer animation
Estimation
Vector space
Fuzzy logic
Element (mathematics)
Moving average
Energy level
Fuzzy logic
World Wide Web Consortium
34:57
Commutative property
Boolean algebra
Multiplication sign
Direction (geometry)
View (database)
Weight
Mereology
Perspective (visual)
Junction (traffic)
Maxima and minima
Mathematics
Mathematics
Logic
Moving average
Ranking
Multimedia
Office suite
Fuzzy logic
Area
Moment (mathematics)
Sound effect
Bit
Price index
Maxima and minima
Degree (graph theory)
Category of being
Resultant
Associative property
Point (geometry)
Event horizon
Field (computer science)
2 (number)
Power (physics)
Social class
Degree (graph theory)
Term (mathematics)
Operator (mathematics)
World Wide Web Consortium
Series (mathematics)
Machine vision
Element (mathematics)
Operator (mathematics)
Line (geometry)
Set (mathematics)
Limit (category theory)
Computer animation
Continuous function
Logic
Personal digital assistant
Query language
Information retrieval
Element (mathematics)
Boundary value problem
Nearring
39:57
Query language
Stochastic differential equation
Degree (graph theory)
Computer animation
Operator (mathematics)
World Wide Web Consortium
44:11
Point (geometry)
Query language
Presentation of a group
Line (geometry)
Multiplication sign
Price index
Mereology
Number
Word
Degree (graph theory)
Term (mathematics)
Database
Moving average
Fuzzy logic
Fuzzy logic
Metropolitan area network
Physical system
World Wide Web Consortium
Area
Information
Physical law
Element (mathematics)
Term (mathematics)
Measurement
System call
Degree (graph theory)
Subject indexing
Stochastic differential equation
Summation
Computer animation
Network topology
Personal digital assistant
Computer science
Fuzzy logic
Element (mathematics)
Data conversion
Representation (politics)
48:42
Statistics
Set (mathematics)
Multiplication sign
Price index
Computational intelligence
Mereology
Weight
Number
Degree (graph theory)
Coefficient
Crosscorrelation
Matrix (mathematics)
Term (mathematics)
Operator (mathematics)
Moving average
Fuzzy logic
Analytic continuation
Subtraction
Metropolitan area network
World Wide Web Consortium
Raw image format
Product (category theory)
Coalition
Theory of relativity
Bit
Set (mathematics)
Term (mathematics)
Similarity (geometry)
Maxima and minima
Subject indexing
Arithmetic mean
Process (computing)
Computer animation
Personal digital assistant
Crosscorrelation
Order (biology)
Statement (computer science)
Element (mathematics)
Fuzzy logic
Coefficient
Communications protocol
Resultant
52:22
Point (geometry)
Musical ensemble
Divisor
Set (mathematics)
Multiplication sign
1 (number)
Price index
Weight
Computational intelligence
Event horizon
Number
Degree (graph theory)
Video game
Matrix (mathematics)
Term (mathematics)
Moving average
Fuzzy logic
Library (computing)
Metropolitan area network
World Wide Web Consortium
Metropolitan area network
Product (category theory)
Term (mathematics)
Maxima and minima
Uniform boundedness principle
Stochastic differential equation
Fraction (mathematics)
Computer animation
Personal digital assistant
Factory (trading post)
Fuzzy logic
Element (mathematics)
Natural language
Freeware
59:04
Query language
State of matter
Multiplication sign
Binary code
Price index
Weight
Computational intelligence
Semantics (computer science)
Information retrieval
Summation
Maxima and minima
Sign (mathematics)
Moving average
Fuzzy logic
Metropolitan area network
Product (category theory)
Gradient
Binary code
Scientific modelling
Sound effect
Term (mathematics)
Discounts and allowances
Degree (graph theory)
Network topology
Fuzzy logic
Identical particles
Resultant
Point (geometry)
Set (mathematics)
Division (mathematics)
Hand fan
Number
Peripheral
Term (mathematics)
Database
Fuzzy logic
Normal (geometry)
Metropolitan area network
Condition number
World Wide Web Consortium
Information
Cellular automaton
Element (mathematics)
Weight
.NET Framework
Word
Computer animation
Search engine (computing)
Lie group
Information retrieval
Element (mathematics)
Abfrageverarbeitung
Family
1:05:20
Point (geometry)
Beat (acoustics)
Set (mathematics)
Multiplication sign
Water vapor
Mereology
Computational intelligence
Distance
Focus (optics)
Formal language
Number
Sign (mathematics)
Degree (graph theory)
Roundness (object)
Causality
Natural number
Logic
Moving average
Fuzzy logic
Subtraction
World Wide Web Consortium
Boss Corporation
Dialect
Focus (optics)
Element (mathematics)
Bit
Line (geometry)
Cartesian coordinate system
Functional (mathematics)
Formal language
Symbol table
Degree (graph theory)
Category of being
Arithmetic mean
Word
Computer animation
Lattice (order)
Personal digital assistant
Logic
Natural number
Interpreter (computing)
Statement (computer science)
Quicksort
1:09:33
Multiplication sign
Control flow
Water vapor
Computational intelligence
Frequency
Mixture model
Database
Moving average
Cuboid
Fuzzy logic
Possibilitätstheorie
Descriptive statistics
Physical system
World Wide Web Consortium
Matching (graph theory)
Key (cryptography)
Water vapor
Theory
Set (mathematics)
Measurement
Maxima and minima
Degree (graph theory)
Word
Message passing
Computer animation
Search engine (computing)
Query language
Interrupt <Informatik>
Freeware
Data type
Reading (process)
1:13:22
Metropolitan area network
Spacetime
Query language
Scientific modelling
Theory
Matching (graph theory)
Information retrieval
Data model
Word
Computer animation
Vector space
Energy level
Maize
Fuzzy logic
World Wide Web Consortium
1:15:38
Point (geometry)
Query language
Musical ensemble
Presentation of a group
INTEGRAL
Multiplication sign
Decision theory
Scientific modelling
Control flow
Price index
Electronic mailing list
Matching (graph theory)
Distance
Number
Moving average
Energy level
Electronic visual display
Quicksort
Metropolitan area network
Physical system
World Wide Web Consortium
Metropolitan area network
Spacetime
Information
Sound effect
Perturbation theory
Term (mathematics)
Measurement
Local Group
Disk readandwrite head
Maxima and minima
Number
Uniform resource locator
Word
Computer animation
Telecommunication
Computer science
Natural language
Energy level
Ranking
Library (computing)
1:17:40
Spacetime
Query language
Point (geometry)
Scientific modelling
Theory
Matching (graph theory)
Distance
Similarity (geometry)
Data model
Information retrieval
Measurement
Computer animation
Vector space
Energy level
Information
Website
Fuzzy logic
Representation (politics)
Abstraction
Data structure
World Wide Web Consortium
1:20:00
Web page
Spacetime
Query language
Multiplication sign
Price index
Distance
Vector space model
Number
Mathematics
Incidence algebra
Term (mathematics)
Vector space
Negative number
Hausdorff dimension
Moving average
World Wide Web Consortium
Axiom of choice
Spacetime
Real number
Point (geometry)
Element (mathematics)
Bit
Incidence algebra
Ring (mathematics)
Term (mathematics)
Vector graphics
Functional (mathematics)
Category of being
Computer animation
Family
1:22:39
Point (geometry)
Set (mathematics)
Real number
Electronic program guide
Distance
Computational intelligence
Mereology
Metric tensor
Trigonometric functions
Mathematics
Sign (mathematics)
Number theory
Symmetry (physics)
Euclidean space
Moving average
Integer
Circle
Office suite
Extension (kinesiology)
Subtraction
Arc (geometry)
Metropolitan area network
Triangle
World Wide Web Consortium
Spacetime
Product (category theory)
Geometry
Basis (linear algebra)
Sound effect
Line (geometry)
System call
Measurement
Distance
Inequality (mathematics)
Similarity (geometry)
Degree (graph theory)
Stochastic differential equation
Arithmetic mean
Computer animation
Circle
Logic
Personal digital assistant
Function (mathematics)
Website
Object (grammar)
Simulation
Identity management
Sinc function
1:25:27
Set (mathematics)
Line (geometry)
Geometry
Time zone
Angle
Vector graphics
Trigonometric functions
Similarity (geometry)
Measurement
Summation
Computer animation
Function (mathematics)
Vector space
Moving average
Addressing mode
World Wide Web Consortium
1:28:13
Point (geometry)
Product (category theory)
Query language
Length
State of matter
Multiplication sign
Scientific modelling
Similarity (geometry)
Binary code
Angle
Distance
Computational intelligence
Number
Sign (mathematics)
Term (mathematics)
Vector space
Utility software
Moving average
Electronic visual display
Normal (geometry)
Subtraction
Physical system
World Wide Web Consortium
Pairwise comparison
Product (category theory)
Scaling (geometry)
Information
Moment (mathematics)
Coordinate system
Term (mathematics)
Trigonometric functions
Vector graphics
Measurement
Similarity (geometry)
Computer animation
Vector space
Personal digital assistant
Query language
Energy level
Data type
1:30:21
Email
Query language
Focus (optics)
Axiom of choice
Theory of relativity
Length
Multiplication sign
Length
Virtual machine
Trigonometric functions
Vector graphics
Functional (mathematics)
Measurement
Distance
Similarity (geometry)
Measurement
Database normalization
Category of being
Computer animation
Euclidean space
Maize
Quicksort
Right angle
World Wide Web Consortium
1:34:18
State observer
Query language
Musical ensemble
Length
State of matter
Multiplication sign
Function (mathematics)
Sphere
Summation
Database normalization
Bit rate
Linker (computing)
Computer configuration
Vector space
Euclidean space
Position operator
Covering space
Spacetime
Sampling (statistics)
Physicalism
Sound effect
Nominal number
Bit
Vector graphics
Trigonometric functions
Measurement
Distance
Maxima and minima
Computer simulation
Network topology
Ranking
Quicksort
Representation (politics)
Resultant
Point (geometry)
Web page
Divisor
Presentation of a group
Computer file
Distance
Rule of inference
Number
Frequency
Term (mathematics)
Representation (politics)
Ranking
Normal (geometry)
Subtraction
Units of measurement
Metropolitan area network
MIDI
Matching (graph theory)
Information
Length
Volume (thermodynamics)
Directory service
Similarity (geometry)
Word
Computer animation
Personal digital assistant
Units of measurement
1:37:11
Spacetime
Query language
Covering space
Sound effect
Measurement
Database normalization
Word
Data model
Dedekind cut
Vector space
Moving average
Divisor
Gamma function
World Wide Web Consortium
Newton's law of universal gravitation
Mobile app
Length
Term (mathematics)
Inclusion map
Number
Positional notation
Frequency
Computer animation
Commodore VIC20
5 (number)
Representation (politics)
Damping
1:42:11
Java applet
Length
Multiplication sign
Scientific modelling
Mereology
Weight
Computational intelligence
Video game
Bit rate
Vector space
Moving average
Electronic visual display
Software framework
Information
Physical system
Logarithm
View (database)
Software developer
Point (geometry)
Basis (linear algebra)
Electronic mailing list
Computer
Sound effect
Term (mathematics)
Measurement
Functional (mathematics)
Demoscene
Category of being
Process (computing)
Frequency
Telecommunication
Hausdorff dimension
Order (biology)
Computer science
Software framework
Quicksort
Representation (politics)
Resultant
Sinc function
Router (computing)
Point (geometry)
Numbering scheme
Momentum
Student's ttest
Distance
Number
Frequency
Latent heat
Degrees of freedom (physics and chemistry)
Term (mathematics)
MiniDisc
Subtraction
Form (programming)
World Wide Web Consortium
Information
Sine
Volume (thermodynamics)
Weight
System call
Power (physics)
Number
Computer animation
Personal digital assistant
Hybrid computer
Information retrieval
Universe (mathematics)
1:45:33
Numbering scheme
Structural load
Logarithm
Inverse element
Weight
Term (mathematics)
Digital object identifier
Number
Frequency
Computer animation
Moving average
Form (programming)
World Wide Web Consortium
1:48:58
Query language
Spacetime
Boolean algebra
Structural load
Run time (program lifecycle phase)
Average
Table (information)
Measurement
Summation
Information retrieval
Vector space
Hausdorff dimension
Moving average
Heuristic
Fuzzy logic
World Wide Web Consortium
Rule of inference
Term (mathematics)
Vector graphics
Power (physics)
Similarity (geometry)
Number
Computer animation
Fuzzy logic
Thumbnail
Data structure
Series (mathematics)
1:51:11
Multiplication sign
Mereology
Independence (probability theory)
Web 2.0
Information retrieval
Vector space
Moving average
Electronic visual display
Relevance feedback
Descriptive statistics
Physical system
Computer font
Spacetime
Real number
Hypothesis
Arithmetic mean
Fuzzy logic
Orthogonality
Resultant
Ocean current
Electronic data processing
Presentation of a group
Algorithm
Set (mathematics)
Similarity (geometry)
Student's ttest
Vector space model
Goodness of fit
Term (mathematics)
Representation (politics)
Ranking
Maize
World Wide Web Consortium
Run time (program lifecycle phase)
Time zone
Polygon mesh
Information
Physical law
Projective plane
Expert system
Independence (probability theory)
Usability
Line (geometry)
Equivalence relation
Subject indexing
Word
Computer animation
Predicate (grammar)
Personal digital assistant
Mixed reality
Library (computing)
1:55:42
Word
Computer animation
Classical physics
Subject indexing
Digitale Videotechnik
Price index
Electronic mailing list
Term (mathematics)
World Wide Web Consortium
1:58:12
Polygon mesh
Scaling (geometry)
Multiplication sign
Projective plane
Price index
Insertion loss
Control flow
Computational intelligence
Subject indexing
Workload
Message passing
Goodness of fit
Computer animation
Linker (computing)
Subject indexing
System programming
Natural number
Moving average
Process (computing)
Resultant
World Wide Web Consortium
2:00:04
Implementation
Collision
Multiplication sign
Mathematical analysis
Vector space model
Field (computer science)
Information retrieval
Subject indexing
Smart card
Moving average
Maize
Implementation
Relevance feedback
Physical system
World Wide Web Consortium
Information
Projective plane
Control flow
Maxima and minima
Subject indexing
Computer animation
Network topology
Natural number
System programming
Figurate number
Physical system
Resultant
2:02:38
Suite (music)
Query language
User interface
Multiplication sign
Range (statistics)
Price index
Computational intelligence
Likelihood function
Information retrieval
Sign (mathematics)
Hypermedia
Smart card
Statistics
Physical system
Cycle (graph theory)
Computer file
Physicalism
3 (number)
Computer
Virtual machine
Right angle
Cycle (graph theory)
Whiteboard
Quicksort
Physical system
Resultant
Row (database)
Point (geometry)
Computer programming
Statistics
Random number generation
Computer file
Variety (linguistics)
Virtual machine
Event horizon
Number
Inverse problem
Subject indexing
Software testing
Data structure
Subtraction
World Wide Web Consortium
Information
Projective plane
Independence (probability theory)
Performance appraisal
Subject indexing
Calculation
Word
Computer animation
Personal digital assistant
Computer hardware
Revision control
Game theory
2:04:42
Spacetime
Scientific modelling
Computer
Theory
Matching (graph theory)
Euler angles
Information retrieval
Data model
Computer animation
Vector space
Smart card
Revision control
Software testing
Information
Energy level
Maize
Fuzzy logic
Library (computing)
Game theory
World Wide Web Consortium
2:06:57
Point (geometry)
Area
Observational study
Independence (probability theory)
Total S.A.
Euler angles
Mereology
Total S.A.
Event horizon
Number
Roundness (object)
Event horizon
Computer animation
Moving average
Game theory
Game theory
Row (database)
2:08:40
Area
Standard deviation
Multiplication sign
Euler angles
Event horizon
Number
Independence (probability theory)
Roundness (object)
Event horizon
Computer animation
Term (mathematics)
Statistics
Game theory
Game theory
Fingerprint
Condition number
World Wide Web Consortium
2:12:10
Area
Context awareness
Information
Bayes, Thomas
Multiplication sign
Euler angles
Event horizon
Thomas Bayes
Event horizon
Computer animation
Theorem
Moving average
Maize
Family
Game theory
World Wide Web Consortium
2:14:42
Information retrieval
Computer animation
Scientific modelling
Computer
Ext functor