Bestand wählen

Elasticsearch Lua

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Erkannte Entitäten
and this is the ones that are you out of the human part in the world in which you have to look at them the stalled shock we got eyes and realizes that if you have a pool of response will have the follows on from plastic and that's classic search building locations on top of ElasticSearch did so good afternoon everyone have my problem 5 minutes of change in year for almost 10 years and I have my passions are distributed systems data analytics and of course so I'm going to talk a little bit about how you can develop bluer plications and use elastic searches a storage or as a base for it so the 1st thing is 1 elastic searches like that HoNOS Elastic Search OK who has use it or is using elastic search but I got a few of you felt ElasticSearch is an open-source document oriented RESTful full-text searching gene and with soft real-time analytics capabilities as you can see so basically it's of Washington the documents are described it as Jason and when you write something to Elastic Search you actually writing adjacent all command that will be able to retrieve exactly as in this at its worst so everything is done over an API on http requests so that's the mean were from Covington and with Jason on an HTTP request you can do almost everything and almost all languages have some support supportable our regarding full-text searching gene you can do all of the things that they you could imagine regarding search so when you're going to make the start writing to the search walks elastic what's running behind it is elastic search and you can implement or completed highlighting e g you features all within the source that you were doing for your website of your application and not only that you can also use the minor which is another open source team from last to view the dashboards with the quarantine that's running behind it that segregating the data is Elastic Search itself OK so it's very colorful so it's very easy to just need to download on and basically execute elastic social you need to having your machine is job up and then using HTTP requests you can do many interesting things and I'm going to show a little bit about it so a last assertion to I was created in the wars Google Summer projects together with flood Luo and elastic as I was in last taken I used to be alive of and the main thing that I started thinking was OK we have lower here and Jason here and want to differences and what do we need to do to make sure that we ride 1 way or we send it in the other way and basically for me they're very similar as of describing documents so for now we don't have any later on the language itself to implement the search or the analytic so basically what you do is instead of writing the Jason you write it in lower and reuse C Jason to change it away so we keep we kept the DSL and all the things that you would write injuries and you just write it in low way we do the change so going to crash basically you have an object to a document the way you wanna college you require the library you instantiate the clients and then you send an index request many want to writing Elastic Search I'm getting an example off of papers and no workshop or the presence of something like that and you have the index which is the bayes I data separation inside Elastic Search and final type and ID for adult men and he had through time defining the paper yeah here the object and sending it with the king name that's recalled by to get you just in the client get again the index the type in the ID that's the real identifier of the document and send elastic you can also do an update so using just the changes that you wanna do so I'm just updating the title over here and of course you can also delete so very simple very straightforward no big deal this year OK the so yeah Elastic Search of course you can you should do search and how do we do that very simple again so index where you want to search the type and the query so here I'm searching for requirements just like that OK there are many internal things are happening and the abstractions that elastic so start to give to you by but this is a nice way of doing searching you get good results as well so again I can send a certain instead of searching to requirements just requirements which will go into all the fields of my documents there actually indexed and searched for this token for this work I can say OK only 1 requirements in the title so I'm buying the word requirements to the title few of my g's and of OK and we can get much much more complex we can actually do so a moon to match so a matching the saying query in multiple fields and actually I'm doing that in many different ways I'm looking to keywords title and the abstract feud but in the words I'm giving a weight of 10 entitling being aware of 5 and an abstract and just giving a weight of 1 that's the deceitful and I'm also adding a few to here which says OK I'm actually just 1 to have the papers for the last 3 years OK so we have this Nice syntax visit on the or the library that you can see now minus 3 1 right what you get back i adjacent response that we transform it into a low object internally than you can handle but which will contain the time to for the requested there was a time of an on the shards this is an internal a piece of that actually contains the data so you can see if you got all the data that you have and then you get the heats which are the actual toll commands that you were getting back to all right so we got only 1 document back and the nexus for for those documents are 0 . 7 in this case and then you get an array with all the hits and for each position of your array you actually get the object itself so here is the object with the index the type and indeed the score for the specific document and then you get the source and the source is x is exactly the same thing that you send to elastic search so the document the jason that you send to Elastic Search the exactly what you get back so you can play around with it OK I could choose I can choose the fuse that I want to but here I'm getting all of them this right OK
but why using ElasticSearch was the motivation behind it why would you start using elastic search and play around with it so I used to
work in this website at pookie in Brazil in the university and this is basically a website with all the papers all the publications for the and you me and Engineering Requirements Workshop all right and it's a very simple website you have like the main page where all the years over here that you can click and look and of search waltz that's base it on Google search for so if you click in 1 of the years you see a raw list of the papers with a few life the number of downloads for the pedia this caller link so you can locate into a scholar yeah and there is also another link that yes you mean to the most cited papers so in order so again in raw form the papers based on the citation so it's a website with 3 pages really simple it was designed it with partially CGI U my sequel and a search was Bazan on the research and why was looking to that and say OK looks old-fashioned to me I want to experiment new things i want to change let's move it into sailor and elastic search so the CGI Luo versus sailor report I leave to a channel probably knows more than me about it but basically I was very easy to that but there is no 1 mountainy CGI more and sailor was the school project with many of new things and a lot of new features and growing in cited why not to try so I just changes everything when Brady's move maybe a topic for another call regarding the database or the data for I had my sequel and now I wanted to use Elastic Search but elastic such is not a database OK you can use it to store data but you don't have transactions you don't have the acid semantics it was not be viewed as a database but do we really need a database or we just need some place to put data and then retrieve data like we update the website year with few documents regarding the papers that were the so lution and that's it I don't need a database for that I don't need nothing fancy with transactions or anything OK so this was good to change on the other side my sequel is not a search engine bad do we really need a search engine simply is you just don't put it on your website Google Source fine in my case it was too early to actually write something here redirect people to a Google page with the links in even worse the results that you get here you cannot change it you cannot do any modifications you cannot say OK titles are more important take downloads into consideration or take citations into consideration is just what Google gives you back and they really wanted to change that so with the last exert could implement my own search result page which you see terrible him looking but I could and I could change the relevancy I could choose if I could give more weight into title abstract dolls I could implement autocomplete highlighting NGO anything I wanted for my application and finally everything would be a new are right
so moving to Elastic Search how was how was it at so we got my
sequel and I needed to go into elastic Search not complicated I the Lewis equal called over here so just install use will rocks read from the database do a select a star iterate over the conferences for each conference so we get all the papers for that conference create what we call the elastic search a book request so instead of inserting 1 paper at a time I got a bunch of papers together and send 1 request with Teddy 15 you can choose the number very simple script took me I don't know 15 to 20 minutes to do that ugly thing all the data is an elastic socially when we change from a
database relational database into tower ElasticSearch which is not actually a database we need to think about OK how are going to model or data to play around with things so we can the normalized data that we can use nested objects and parent and child all of them all complex but I chose parent and child just to keep exactly what I we can do 1 to any relationship here and that's what I wanted 1 conference many papers so How was the mappings or the schema colored mappings so the 1st thing it's ElasticSearch has dynamic mapping so you just need to send documents in to look at the document and figure out an updated mapping accordingly but sometimes you wanted to weaken so I needed to tell Elastic search that papers have apparent which conferences so it was creating a relationship here and it's a that idea was an integer in year was also an integer or otherwise is a string I would get problems doing sorting the and I also needed to tell the paper sessions should not be analyzed it this is good when you want to do sorting as well because if you analyze a string you get problems I don't have time to get into that but that's the simple configuration had to do I didn't have to tell about title or any of the few all of the other fields if they strings they would be analyze it and available for search then the
local changes loading almost the same instead of select star ideas needed to write a little bit different I want to sort by ID I want all of the documents this is a number actually choose to be a local variable like 1000 send the query in here again is just the we get instead of select we're just changing a little bit the way right it's very straightforward again of course modify the
search balls from the Google into the search box and now instead of this page where actually I get back my own main roof thing and also here I can only get a list of all the documents I had so I'm searching for it is a close which means requirements in Portuguese I can't even tell the user about the quality of the results and I can do this fate I want I'm not good with you wiser just left it from anyway you could to make it the How was
the search simple function Querry size the index the type the query here is exactly this name so that's
what I'm passing here I send a search I get that results I put it in the screen OK is it that simple to do the search yes is you can do it much more complex if you want but I just wanted to prove my point I finally Elastic is
not alone so when you going to be elastic family you have a lot of things to play around
so we have to wanna as 1 of the most interesting and so I could actually Estelle beats and start collecting the data from the computer like top information CPU RAM just dump it into lasting search and then you denies this ward like that so I can see the papers over time papers in English Portuguese and Spanish the most of keywords that appear in the documents and any metrics that you want to put together the so the nexus that's well we need more use cases more real applications we need to improve tasks in the library actually update to the new versions Elastic Search ultimate integration and finally when is mature enough make it unofficial client and of course the dual rocks rocks back yeah I take you were much for listening if you have any questions I think we have 4 minutes few who the following fare for questions the 1 question questions charge people that actually such the
Desintegration <Mathematik>
Atomarität <Informatik>
Formale Sprache
Familie <Mathematik>
Fortsetzung <Mathematik>
Kartesische Koordinaten
Formale Semantik
Prozess <Informatik>
Trennschärfe <Statistik>
Skript <Programm>
Figurierte Zahl
Inklusion <Mathematik>
Lineares Funktional
Gruppe <Mathematik>
Gebäude <Mathematik>
Güte der Anpassung
Klassische Physik
Arithmetisches Mittel
Rechter Winkel
Ganze Zahl
Automatische Indexierung
Projektive Ebene
Ordnung <Mathematik>
Lesen <Datenverarbeitung>
Web Site
Gewicht <Mathematik>
Gebäude <Mathematik>
Analytische Menge
Open Source
Virtuelle Maschine
Endogene Variable
Turm <Mathematik>
Elastische Deformation
Relationale Datenbank
Open Source
Physikalisches System
Binder <Informatik>
Mapping <Computergraphik>
Objekt <Kategorie>
Wort <Informatik>


Formale Metadaten

Titel Elasticsearch Lua
Serientitel FOSDEM 2016
Teil 93
Anzahl der Teile 110
Autor Musa, Pablo
Lizenz CC-Namensnennung 2.0 Belgien:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen.
DOI 10.5446/31008
Herausgeber FOSDEM VZW
Erscheinungsjahr 2016
Sprache Englisch

Inhaltliche Metadaten

Fachgebiet Informatik

Ähnliche Filme