Merken

Syntax isn't everything: NLP for Rubyists

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
this work and home lives and it so never 1 sees green because and then tried to know and don't know because you know what's live on any will images so I can really ironic itemize most places on the internet measures taken on winter so it's not atomizer I really getting people to any questions comments you completely wrong about that type things during talks and my phone is somewhere over there and which literature turned off on top so Yale won't see it anyway so feel free doughnuts and long once a week this week skin real challenging that that was upon in various context demobs the the speaking all sorts of stuff an article on cloud platform so they and cloud community is other things we do machine learning in the eyes I am happy to answer questions I also have a plethora of opinions that I'm more than happy to share with you I mean confined will chat and some support for a large company omega cast to saying that any code this
talk is copyright along licensed Apache detail as called is cannot currently on get home on but it will go you an X near 0 and I'll post about on my blog when it gets there are posted on Twitter so
hopefully this since the room is relatively empty people are just randomly here because they didn't like the last talk they went to or in a locked in Willie but what is unknown and all the i gotta why did you use letter penalty is natural language processing which which gave you a bit closer to the idea behind my talk but still of fuzzy for me sort the intermediate the natural language
processing is the field of computer science artificial intelligence and computational linguistics concerned with the interactions between computers and human life natural languages in particular concerned with programming computers to fruitfully process large data corpus so that is 1 exceptionally long sentences on I winding words so here's the definition I use teaching computers to understand and ideally respond to human languages and regular languages and in things like English Japanese American Sign Language British Sign Language all things like that languages humans use so how to echo millions of middle school is everywhere wanna use this such why should I care and the reason is a battle being is already here who and interact with a phone system or they're like say automation number and the standard layout and did you say again I have a restraining a 1 in parking on 11 Khamenei to make sure player was originally 3 we can monitor in your on a magical was window pops up and it's like this this embodied person and the sentence if you need help you help you annually cognition is really a personal and not have happened last week it was pretty am in on all these things natural-language-processing of some sort for going bad and all the is likely involved but the promise among is actually better user experiences I wanna live in a world were instead of having to teach people how to interact with computers we can teach computers how to interact in ways that people already interact successfully and so example of non ideal NLP in 1 many of you are probably familiar with is this my this is my favorite sliding into the of computer T. Earl Grey hot so that is an example of pattern the where 4 is a request is very specific to what I want to have happen I will use that phrase with any other person and move by when I was on the that seemed really futuristic it turns out we can do better than already of these random things the linear non-sentential ordered all has a story or other things are a lot of people's houses and on great topic John and Angeline yesterday where they were arguing with multiple axes appear on the stage I have a I really like it and depending on the particular brand of these they have a relatively large set in some cases and nearly open completely leaderless corpus of voice commands they can respond to and if you really want to me as of both were roughly 1 of these office with 30 tomorrow and a lot of formal show it is to ground to the end of the talk and and then we also I
mentioned before have tech support inventories trees and these are actually getting a lot better in my experience I can actually call my credit card company and say I not like God and they will send me to a right person into the question for me and I really hope that this stuff gets parameter next couple years but arsonist cases you may not have thought about accessibility and the closing keynote
yesterday of leaking out 1 of the points that was made was that children often use voice interfaces because they can't read yeah I used to work on software for kids and was presented with specializing hidden learning disabilities so the ability to use voice interaction and not have to type and spell correctly is fantastic for people with dyslexia or perhaps people of a broken arm and content consistently right now or maybe they're holding a baby on there's all sorts of use cases where this kind accessibility matters in addition the stuff that many of you are probably thinking when I put the word accessibility on the slide which is blind people blind people can use voice interaction and the other thing that if NLP can help us with is not the same from our understanding and our ability to analyse large amounts of data so what's going happen as a feedback button somewhere it so we worked intact for kids the stuff we got the feedback money was amazing neurons had the best way is telling you may hate your software so I know really it was lost but once you have more popular processing on the feedback is really hard initially everyone not everything back you know that came in and around us has to go to a folder and we want to build like brown being back to the right people like if it was from adult about billing that should go to the team that you don't processing and set prices and stuff it was again saying you were the head and ii this unique things to read on a bad day of older and it was like I don't understand what's new new and teams working on that interface to try to figure out we can make the instructions butter but we going do that it was taken 1 person going through flagging all that's is done by the and how they can also be used to assist us in other ways I want recorded tool called the rat that analyzes your e-mails Gmail pondered and if you try to send an e-mail it comes off as hostile it tells you you might wanna rethink that imagine how many get home flame wars could be stopped is everyone had that thing so why don't we haven't already the because energy is for like really really hard enough to leave me wires something on in English is horrible so pop quiz this is a word this word is sealed everyone imagine we think of when you think of steel the whodunits
Ch this nonetheless who got a musician who thought of
something completely different so 1 word at least 4 different things that some people in this room I thought of it is actually fairly evenly distributed whose conical because when I ran this was some other people everyone not that can
in and new like the picture which is why want the word seal and then we obtain is known homophone
words but yeah there's so many English teachers so little time the and then we have words they can be multiple parts of speech for example Lumpkin numeric she loves her white can also be now 1 last forever so I repeat In horrible and even and stuff like a regular verbs slang idioms and all the other bits that make a language a human language and make learning a foreign language we we Norway and it turns out that English is unique to a particular moment in which is especially horrible but all human languages are actually pretty horrible and they're all horrible in the way that discussed on different ones even the ones we're trying to manufacture to be less formal or still horrible because humans make idioms humans makes line language evolves and particularly formal for computers because there is no formal closed grammar for hanging with him languages some languages model and others but if you think about the grammar for computer language for glycidyl appeared to print and text the general are regular there's a limited vocabulary and there is a limited vocabulary have like this can be a symbol symbol can start with the following characters can can contain the other than you can contain spaces if you don't understand the words formal close grammar let's compare to chat with called drag Chomsky Italy also afterwards also the alignment horrible because humans are really bad it being
precise for example if I say
I'm starting there's a remote chance and that's true this is probably not and if I say someone you look freezing again remote chance it's true I hope not as it's probably not true we exaggerate a lot I was reading a whole article doing research for this talk that were unique is becoming less unique over the last 30 years 30 years ago newspaper articles unique is relatively rare frequently editors and reporters would use unusual which is the word that probably won't notice that now I think it's a new almost as likely to have more unique as you are to have unusual when you need unusual I unique is something like 7 times more common than it was in print articles than it had some years ago so and is not an example of language evolves rather favorite examples literally because as a way to years ago doesn't mean literal literally anymore so and there's also the problem of computers
sockets sarcasm like neutrally and the bottom it's not just humans society communities active this example is the previous talk so I can say sure and you want to help you out your problem I can also say sure I'd love to do the dishes depending on how I say that the meaning changes and also changes by the context words that are around it and despite what we learned from hitchhikers guide to the galaxy computers are really bad at distinguishing sarcasm and the even worse at generating ieee somehow integrative they so as you why is this hard problem confused you and this chair but the big thing is is that English is hard and natural language processing is hard and languages are hard because humans I mean your many of your arguments use language in where weights we use sarcasm is exaggeration and it's hard because human languages are complicated and always changing but since created human language we can simplify this whole thing to why is it hard to natural image processing is hard
because units now joint harder passing is hard because people we are complicated and we make things hard for computers so hopefully dimension this is hard but not all is lost because we are
working on this problem for a really long time what these 2 famous mathematicians were similar spin talking on natural image processing they proposed do on translation between languages and I am of the generation that built translate became thing while I was conscious and somewhat through some useful for me and I've seen automated translation improved greatly Over the last 15 to 20 years but we're working on this problem for several hundred years at this point another example on 1950 Alan Turing proposed during test this well the terrorists use or it is a couple the moderate and so I mean actually the the definition because a lot of people think they understand it and they don't of including the regionally so the Turing test as a way to test for machine intelligence and uses natural language a judge an impartial judge what is a conversation between a human and a machine without knowing which 1 is which usually this means the text going across the screen the judge knows that 1 of the 2 is a machine but as a which 1 is the machine if the judge can't tell which is the human and which is the machine the machine passes the Turing test ends people at trying so that's 1 my favorite examples is otherwise this came out shortly after the just actually and if your Enoch search you can do that x doctor allegedly with in real time and because I have the time we
get to that right now I wasn't
so this is the unit psychotherapist sums a high I'm feeling son and says hi honey and you know what brings you to me and say told 0 and viewing sad why do you say your article manager feeling sad because I already told you the stupid machines the is it because your article names Mr. machine became yes well I see what makes you believe this is so a so this is a very classic and example of LT there's lots of limitations and this line project for the on is to pour a Python implementation remain because why not you have to go to a conference and that in some really stupid hallway programming process but it's also really fun and only some users 1 my favorite current projects are for working with middle schoolers they like being snarky they like you know talking back in making things a back based on what you say is a lot of fun such my 1 of my examples like there we get also analyzes action classic example of a chat so you're in the jackpot I have a lot of fun you do those but the fact that chat bots have been around since the late fifties early sixties kind of surprised me I thought we invented that stuff when I she turns out no so we didn't it's there on forever some background on for about 15 minutes and to a single line of code and that's not really fair made so current the everything and I'm ready talks elated super practical examples that hopefully no 1 will actually use idea who is the idea instead and Conf 2015 I give a talk this this was the talk
I was stupid ideas from many changes in I'm talking demonstrated how do sentiment analysis of tweets by scoring indian logion no they contained on sentiment announces a
subfield of the image processing and the gold analysis to figure out of a given body of text is generally positive general leaning in or something else and using what you do this they give common ontology score is based on positive negative they work this was my general
scale on non-polynomial toxic put the privilege of the screaming many times and I used energy at the time and you go back and watch the video on concrete because NLP is stupid hard and in November 2015 I don't have the skills or the ability to train upon model did you accurate natural language processing on all the crazy so the people tweet during a conference in the army in real-time but to about that most folks who do machine learning of any kind I I don't actually build their own models in building models is hard takes a lot of time takes a lot of knowledge about various ways of making a wrong and last year tons and tons of pretrained models that you can activate the out of the and I'm using this 1 all during this class natural language anomalies that unwanted today instead of my have emerges scoring scheme so we have jam is actually supposed to be dash between cloud languages are not that dumb installed will come language I believe it's all maybe data done gene's all of them haven't underhand Sullivan and here's all the colony and do the scoring and was method not required think creating new object will analyze method and create a document using the trait at the text I call the sentiment method and it gives me an object that has scorned magnitude chorus how positive or negative it has negative 101 manager is how much that thing in itself you know I get a score of 0 . 1 in magnitude of thought and that is a very very neutral tweets invite warning line and a magnitude of 0 . 1 it's not have with not being excessively negative like there's been some of the remaining and 1 post about how to interpret those if you're curious is understanding those 2 things a little bit of a pulp don't multiply them together it doesn't work that way I tried that so i'm gonna massively him ways over how I set this so let us into communities cluster and got MapReduce system with something that's pulling data from Twitter I don't have been exploited afterwards come the birthrate show all running I explain it all in great detail in my original talkers Mariel toppers Henderson and distributed systems to do crazy things and I'm going to do and that now at the end of the talk if we have time but I need your help and stuff from that hashtags of was the phone or something and make some sort of tweet it can you know blame we talk at some point in its weight and tried in a sentiment but generally sends run been running pretty good donor conference as far is not surprising how much you drink of water yelled treat something all it if the we get a eyeballs thinking training on talking is an expert is not me telling you about grammar for minute it's in that so I was talking about was playing with it like have they will serve diagram sentences in grade school is I totally then I hated every minute of like doing this thing over and over and over again during lots never seen these before this is 1 form of sentence diagram In subjects vertical line number have vertical line direct object in all the stuff the modify the was this the lines so I will start items like this is under his love I talked about other people engages programs look more like this on involvement in the diagram in the verb is at the center and everything that modifies it goes off to the sides and also all words organized so that they're connected to the word that they'd modified and when I started showing this to some references LIB their light you grammar I hate you and then like so you know what what wonder adopted like what you remember and current so brief side quest
if grammar I am apologize and installed non-native English speakers on the on and you all know this already gives you've all had to study this interior on back and government so
I 1 way we understand words is read by labeling them based on the function is called parts of speech words were actions you can have a sentence number 1 kind of action might be
jump and you can also firms that are
active state of being words like the and we have nouns
what was learned now known as a person place a bank that is strictly true a person like Max around tearing a place like the
bathroom or Phoenix a thing
like a cactus on but you can also
ideas are you may have heard the phrase abstract now democracy love
those are abstract mountains concept it in the way we break words
adjectives and described or modify other words usually nouns this great podcast explains why that is strictly true because again humans are on edge of its and since visibility activated things bloom small 5 mils are all adjectives they can also compare things near and far and if you are of the same generation and your thinking about a sketch from Sesame Street when I say those words almost invented it but I didn't because copyrights of but please go search search near and far Sesame Street on a video search its whole area and we have things called the islanders articles a handout but modern grammar calls them determiners and help clarify which nouns and determiners also include words like this and that all articles or determiners not all determiners articles it's like squares and rectangles it's and we also part of a sentence which are different parts of speech so the road is the thing you need to have a sentence union mom therefore the is the verb the subject is the an element of work it and the direct object is the thing that ever happened to most likely now all you need to know the populace here's a sentence the captain's such the subject is kept the verb is it's the direct object is fish Frank was complete and all standard
engine and now that was you reminded how grammar English works the so we to working out sentence
diagramming here is the basic idea of the sentence diagrams subject verb direct object other stuff and I want to go units and to do this I need to know what part each of these were dissolving well known to assume that it and so is the sentiment of a noun using the same textbook and it returns a list of tokens and turns way more information about every single token than you would ever want this that I would want the time of here this is the topic of the word cat in the sentence the cat eats fish the so here is the text itself and where it appears in the sentence of the offset here's the part of speech is a noun in singular I think this is actually have a lot of modifiers announced we Anthony's grammatical case is going on and on and grammatical gender all in English which can pull on grammatical mood and tense tense is generally known upon announcing list and all the information is available if it's relevant you use this on something like German or Spanish or something that has more cases more jet of grammatical gender things like that and see how that works on and then this talk in his the enabled in En subj for subject and so what this is saying is that cat and sentence cat so the cat eats fish is the subject of sentence so with this I have enough to make myself some also ask your diagrams here and this finding whatever token as the label subject taking the text in the same thing defined Lemma labeled where called that number ask your and I have some absolutely amazing are area here of 1 cool thing if you haven't seen it before knowing neuroblasts you can multiply a string by a numeric it has to be in the order you can't switch it but I'm multiplying a space by the language by the length of the words The subject would here so when everything lines up correctly so also without the direct object defined the things label direct object even better more rock and ASCII art thought how easy but I'm missing a word anything I have looked fits so the natural image priors in 1 other useful thing it is nothing called had token index for each word and this is the index of the parent word in a corpus of text for the current node and not parent word parent token but most of the tokens are words you'll see solid not always so this is the token from of its head OK intent open index is 1 that's a list of tokens so what this is saying is that gonna refers to cat everyone follow navigable tricky so just going go through all my tokens and find everything refers to the subject and we have the same and token index as an index of my subject of print that out and and you know well more insecure I have a basic sentence diagramming I couldn't figure out how do diagonals ASCII art and pretty OK with that as it turns out this is often this the problem resulting attorneys on something about a super happy about this I in artisans like this on the and then go so well as you can probably imagine so it is that that and talk to some reference about why Indians and diagram this way so I fell back on my my favorite tools everyone's got their favorite tools set and monitor tools is a jumbled graph this is actually generated my very 1st conference talk about in a long time ago I and all that does is the geminates creating node node and graphs not charts like bar charts graphs like now rests on easy in provides a simple DSL and it creates doc files and programs on graph is really not files and builds visualisations Fourier and all you need to know for graph is and has 2 methods inside of note which takes 2 optional arguments and required ID and optionally more and edge which takes an idea for the 2 in the the units and you wanna look back and so binding and this is all a company to build a based sentence diagram this on a Harrison got boilerplate is just as in this block media so applies a digraph because it is a directed graph you that it also I is that index as the idea in the text of the label free to my nodes making up for every token I'm going to making ends from the current node to its head yes it doesn't refer to itself and in below back sky confusing they most known applied so that's the cat fair expression alone is that actually have the punctuation punctuation is considered the tokens well and then that's more complicated sentence and then we can always my 1st go the cat means that it's not this with a side of milk using a prepositional phrase there because milk with the milk 10 tokens of because milk is the end of the prepositional phrases of the beginning of prepositional phrases and grammar is also so that governments left on as citizens so examples this is what I really like to do any of these talks because of ethnic serious call the time I get bored but there also to practical uses and we talked about customer feedback we talked about some talked about making ways to make our products more usable for a wider variety of people and I'm hoping that some of you have ideas of your old at this moment and hopefully that my job and if you want to start I wanna play around this on the go matrilineage API is going to play around you that the 1st 5 thousand requests to each endpoint syntax sentiment minimum called entities all of the factors of the 1st 5 thousand 133 and its price per thousand from that and if you we will the present but it's very very reasonable and just because I might do this is so much fun I ran job walking through it and it got a syntax analysis correct as a woman surprised by that so in no it's a of around experiment technologies can see it's it's worth like I highly encourage you display because we learn all these new concepts might binding and so what should you will where else config medical labs and stickers energies are pale sometimes restricted appear depending on your parents I we have talked this morning called go problems remains interact during this that's a lot on contracts and when record is doing a talk on instrumentation what map is really doing in production tomorrow at 3 30 in the same room and as a similar were giving a rat giving identical home there's no link it's also Andreas I wrote it on the web were on talk with myself so this is what is it thank you ask
if you have any questions on that 30 minutes exactly stay silent invertible questions yes so the I tried it so the question was how does this understanding approach to deal with current grammar and tried it and unreasonable I tried was on bodies heart and try out the period and it couldn't figure out the bunnies was the subject I have a deterrent figured it out so it takes its best guess is a machine machine make mistakes and it's getting better and all these models improved over time they get better and better for the most part it's pretty darned accurate my grammar is in general is really bad and I have a copy of it from a blog so just to make sure they don't do horrible things mostly 2 or more things almost on and it's generally pretty known for especially from the from common mistakes that people make that means for improperly using something called a verbal using Collins less common mistakes it's not as good the the training so that was has movies plant disease that empowerment because it is a simple and actually no II II and running the center and promised you guys the closing the demo here I'll show you so this
is my so this is running in real time and you may not be able to see that all you have so next here
when we hit the button so that you can search string theory much Sauternes sentiment is 57 so even if you the training 1 mole while you out you outweighed by the positive because it's certainly positive and there's no reason to its and thus far have assumed effect and significantly way but I don't know what you don't know the science behind on another modeling model to build and to act some doesn't seem to complain the ruin it I know that so Mr. writer question of the summary right How does cleaning up data ensuring and higher quality data that improve this syntax analysis we then at some also I actually really lazy and I haven't done any cleaning of data because I'm a tester part and I like to try to break things and any time we have to put additional our preprocessing and that's slowing me now on I can't imagine that it would hurt but I'm pretty happy with what it's done without on Net entity analysis is actually pretty good without I through sometimes it it's time but through I love the con pilot light thing for entity analysis and divide and conquer by ice cream were the 2 most important and is that sentence and I agree with that sentiment has so it is generally it's announces generally is freedom find proper nouns cities companies things like that a 1 of somebody I know I used to work with does language processing of an SEC filings I'm looking at various things trying understand you know what this company sang you cannot big-name companies and stuff in cities and things like that those separate things but you president of become part screen components could pretty analysis that so other questions for any other sources of error correcting university you're not photometry share the so if at
the a ge a
Streuungsdiagramm
Maschinencode
Datentyp
Systemplattform
Algorithmische Lerntheorie
Kontextbezogenes System
Cloud Computing
Bildgebendes Verfahren
Quick-Sort
Auswahlverfahren
Computeranimation
Internetworking
Prozess <Physik>
Prozess <Informatik>
Web log
Twitter <Softwareplattform>
Natürliche Sprache
Maschinencode
Natürliche Sprache
Computeranimation
Prozess <Physik>
Formale Grammatik
Abgeschlossene Menge
Zahlenbereich
Interaktives Fernsehen
Computer
Computer
Computerunterstütztes Verfahren
Reguläre Sprache
Computeranimation
Netzwerktopologie
Datenverarbeitungssystem
Vorzeichen <Mathematik>
Natürliche Sprache
Bildschirmfenster
Mustersprache
Optimierung
Informatik
Hilfesystem
Parametersystem
Prozess <Informatik>
Ideal <Mathematik>
Störungstheorie
Physikalisches System
Optimierung
Natürliche Sprache
Quick-Sort
Chipkarte
Office-Paket
Datenfeld
Informationsverarbeitung
Sprachverarbeitung
Wort <Informatik>
Computerlinguistik
Neuronales Netz
Rückkopplung
Addition
Punkt
Prozess <Physik>
Interaktives Fernsehen
NP-hartes Problem
Quick-Sort
Computeranimation
Rechenschieber
Energiedichte
Menge
Software
Datentyp
Wort <Informatik>
E-Mail
Schnittstelle
Schreib-Lese-Kopf
Wort <Informatik>
Formale Grammatik
Programmiersprache
Bit
Momentenproblem
Formale Grammatik
Abgeschlossene Menge
Sprachsynthese
Symboltabelle
Computerunterstütztes Verfahren
Natürliche Sprache
Computersimulation
Raum-Zeit
Computeranimation
Eins
Multiplikation
Mereologie
Wort <Informatik>
Luftreibung
Computersimulation
Gerade
Parametersystem
Prozess <Physik>
Gewicht <Mathematik>
Natürliche Zahl
Mathematisierung
Hochdruck
Gefrieren
NP-hartes Problem
Bildanalyse
Bericht <Informatik>
Computerunterstütztes Verfahren
Kontextbezogenes System
Natürliche Sprache
Socket-Schnittstelle
Computeranimation
Arithmetisches Mittel
Texteditor
RPC
Wort <Informatik>
Eindeutigkeit
Gefrieren
Softwaretest
Umsetzung <Informatik>
Punkt
Hausdorff-Dimension
Natürliche Zahl
Bildanalyse
Computerunterstütztes Verfahren
Natürliche Sprache
Computeranimation
Virtuelle Maschine
Generator <Informatik>
Softwaretest
Einheit <Mathematik>
Turing-Test
Mathematikerin
Translation <Mathematik>
Message-Passing
Touchscreen
Maschinencode
Prozess <Physik>
Gewichtete Summe
Gruppenoperation
Klassische Physik
Einfache Genauigkeit
Implementierung
Computeranimation
Chatbot
Virtuelle Maschine
Einheit <Mathematik>
Datenmanagement
Rechter Winkel
Inverser Limes
Projektive Ebene
Optimierung
Gerade
Prozess <Informatik>
Euler-Winkel
Ontologie <Wissensverarbeitung>
Twitter <Softwareplattform>
Mathematisierung
Bildanalyse
Zerfällungskörper
Computeranimation
Analysis
Bit
Gewicht <Mathematik>
Wellenpaket
Prozess <Physik>
Punkt
Wellenlehre
Wasserdampftafel
Klasse <Mathematik>
Formale Grammatik
Zahlenbereich
Computeranimation
Gradient
Videokonferenz
Knotenmenge
Bildschirmmaske
Datenmanagement
Natürliche Sprache
Algorithmische Lerntheorie
Optimierung
Hilfesystem
Computersimulation
Gerade
Streuungsdiagramm
Zentrische Streckung
Expertensystem
Güte der Anpassung
Gebäude <Mathematik>
Nummerung
Physikalisches System
Natürliche Sprache
Quick-Sort
Programmfehler
Objekt <Kategorie>
Energiedichte
Diagramm
Echtzeitsystem
Twitter <Softwareplattform>
Demoszene <Programmierung>
Server
Wort <Informatik>
Größenordnung
Innerer Punkt
Streuungsdiagramm
Lineares Funktional
Wechselsprung
Mereologie
Mereologie
Gruppenoperation
Sprachsynthese
Zahlenbereich
Sprachsynthese
Wort <Informatik>
Aggregatzustand
Computeranimation
Aggregatzustand
Subtraktion
Mereologie
Abstraktionsebene
Formale Grammatik
Rechteck
Paarvergleich
Sprachsynthese
Element <Mathematik>
Fastring
Computeranimation
Videokonferenz
Objekt <Kategorie>
Generator <Informatik>
Quadratzahl
Flächeninhalt
Mereologie
Determiniertheit <Informatik>
Attributierte Grammatik
Wort <Informatik>
Hilfesystem
Standardabweichung
Promille
Momentenproblem
Extrempunkt
Natürliche Zahl
Hochdruck
Formale Grammatik
Ungerichteter Graph
Raum-Zeit
Gerichteter Graph
Computeranimation
Freeware
Arithmetischer Ausdruck
Einheit <Mathematik>
Prozess <Informatik>
Natürliche Sprache
Maschinencode
Momentenproblem
Nichtunterscheidbarkeit
Visualisierung
Figurierte Zahl
Gerade
Parametersystem
Dicke
Systemaufruf
p-Block
Biprodukt
Natürliche Sprache
Teilbarkeit
Token-Ring
Automatische Indexierung
Geschlecht <Mathematik>
Information
Ordnung <Mathematik>
Computerunterstützte Übersetzung
Diagonale <Geometrie>
Zeichenkette
Varietät <Mathematik>
Fitnessfunktion
Rückkopplung
Gerichteter Graph
Zahlenbereich
Sprachsynthese
Graph
Benutzerbeteiligung
Knotenmenge
Datensatz
Vererbungshierarchie
Optimierung
Konfigurationsraum
Bildgebendes Verfahren
Schreib-Lese-Kopf
Streuungsdiagramm
Gerichtete Menge
Graph
Geschlecht <Mathematik>
Token-Ring
Mailing-Liste
Binder <Informatik>
Elektronische Publikation
Parser
Design by Contract
Mapping <Computergraphik>
Objekt <Kategorie>
Energiedichte
Diagramm
Flächeninhalt
Offene Menge
Mereologie
Hypermedia
Wort <Informatik>
Textbaustein
Demoszene <Programmierung>
Virtuelle Maschine
Echtzeitsystem
Wellenpaket
Web log
Mereologie
Formale Grammatik
Frequenz
Figurierte Zahl
Computersimulation
Computeranimation
Prozess <Physik>
Wellenpaket
Gemeinsamer Speicher
Rechnernetz
Computeranimation
Zusammenhängender Graph
Grundraum
Computersimulation
Analysis
Touchscreen
Superstringtheorie
Streuungsdiagramm
Softwaretest
Soundverarbeitung
Präprozessor
Digitale Spaltung
Systemplattform
Quellcode
Elektronische Publikation
Parser
Natürliche Sprache
Rechter Winkel
Mereologie
Eigentliche Abbildung
Streaming <Kommunikationstechnik>
Zentraleinheit
Fehlermeldung
Datenverarbeitungssystem

Metadaten

Formale Metadaten

Titel Syntax isn't everything: NLP for Rubyists
Serientitel RailsConf 2017
Teil 45
Anzahl der Teile 86
Autor Hammerly, Aja
Lizenz CC-Namensnennung - Weitergabe unter gleichen Bedingungen 3.0 Unported:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen und nicht-kommerziellen 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 und das Werk bzw. diesen Inhalt auch in veränderter Form nur unter den Bedingungen dieser Lizenz weitergeben.
DOI 10.5446/31286
Herausgeber Confreaks, LLC
Erscheinungsjahr 2017
Sprache Englisch

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract Natural Language Processing is an interesting field of computing. The way humans use language is nuanced and deeply context sensitive. For example, the word work can be both a noun and a verb. This talk will give an introduction to the field of NLP using Ruby. There will be demonstrations of how computers fail and succeed at human language. You'll leave the presentation with an understanding of both the challenges and the possibilities of NLP and some tools for getting started with it.

Ähnliche Filme

Loading...