Teadmiste formaliseerimine 2019 - Redigeerimiste ajalugu

http://www.lambda.ee/w/index.php?action=history&feed=atom&title=Teadmiste_formaliseerimine_2019 Teadmiste formaliseerimine 2019 - Redigeerimiste ajalugu 2026-11-07T00:58:15Z Selle lehekülje redigeerimiste ajalugu MediaWiki 1.43.11 http://www.lambda.ee/w/index.php?title=Teadmiste_formaliseerimine_2019&diff=16537&oldid=prev Tammet: Uus lehekülg: '<table> <tr> <td> teadmised </td> <td> Name: <b>Knowlege representation</b><br> Code: <b>ITI8700</b><br><!-- Link: http://www.lambda.ee/i...' 2020-01-29T14:24:37Z

<p>Uus lehekülg: '<table> <tr> <td> <a href="/wiki/Fail:Mephistophiles_small.gif" title="Fail:Mephistophiles small.gif"> left|teadmised</a> </td> <td> Name: <b>Knowlege representation</b><br> Code: <b>ITI8700</b><br><br /> Lecturer: Tanel Tammet<br><br /> Labs: Evelin Halling and Tanel Tammet<br><br /> Contact: tanel.tammet@ttu.ee, 6203457, TTÜ ICT-426<br><br /> Archives of previous years: [[Teadmiste otsing, formaliseerimine ja hoidmine|2015]]<br /> [[itv0060 2014 arhiiv|2014]], [[itv0060 2013 arhiiv|2013]], [[itv0060 2012 arhiiv|2012]], [[itv0060 2011 arhiiv|2011]], [[itv0060 20010 arhiiv|2010]], [[itv0060 2009 arhiiv|2009]], [[itv0060 2008 arhiiv|2008]], [[itv0060 2007 arhiiv|2007]], [[itv0060 2006 arhiiv|2006]], [http://cs.ttu.ee/kursused/wav4130/ older].<br /> </td><br /> </tr><br /> </table><br /> <br /> == <font color="red">NB! This is an archive from 2019, not actual course contents</font> ==<br /> <br /> == <font color="red">Exam results</font> ==<br /> <br /> Siin on <b>lõpptulemus</b> kolmanda eksami järel: [[Teadmiste formaliseerimine lõpptulemus 2019]] koos praktikumipunktide ja hindega. <br /> <br /> Milles seisnes vahepealne parandus: praktikumipunktide kokkuliitmisel oli kaks erinevat interpretatsiooni: kas kolmas praks on osa summast 0..40 või on ta puhas lisand esimese kahe praksi summale. Esialgne hinne tulenes esimesest interpretatsioonist, parandatud hinne teisest. Selle tulemusena mitmed hinded tõusid.<br /> <br /> <br /> <br /> Viimane eksam toimub kolmapäeval, 12. juunil kell 12:00 IT majas ruumis ICT-A2.<br /> <br /> Siin on esimese eksami tulemused (praktikumipunkte ei ole lisatud):<br /> [[Teadmiste formaliseerimine esimene eksam 2019]]<br /> <br /> Kellel alla 30 punkti, palun eksam uuesti teha. Igaühel on võimalik teha eksamit kaks korda.<br /> <br /> == Exam ==<br /> <br /> All exam times on Wednesdays at 12:00 (lecture timeslot):<br /> <br /> * 22 May 12:00 in the economics building SOC-211A<br /> * 29 May 12:00 in the economics building SOC-213<br /> * 12 June 12:00 in the IT building ICT-A2<br /> <br /> Exam is in written form, no materials can be used.<br /> Exam will last 3 hours, but most likely you can finish in ca 1.5 hours.<br /> <br /> Here are notes as [[What will be asked in the ITI8700 exam]] including materials to study.<br /> <br /> <br /> <br /> <br /> <br /> <br /> == Time, place, result ==<br /> <br /> Lectures: Wednesdays 12:00-13:30 room U06A-229<br><br /> Practical work: Wednesdays 14:00-15:30, room ICT-121, ICT-122<br><br /> Practical work will give 40% and exam 60% of points underlying the final grade. <br /> The exam will consist of several small excercises.<br /> <br /> The first practical work time on 30. January at 14:00 will be used as a secord conventional lecture of the day.<br /> <br /> Some of the last lecture times at the end of the course may be used as additional seminars/labs.<br /> <br /> == Assumed background ==<br /> <br /> You should have studied the course [http://lambda.ee/w/index.php?title=Tehisintellekti_ja_masin%C3%B5ppe_alused&oldid=15188 "Basics of AI and Machine Learning"] or get acquinted with the logic and probabilities parts of the course yourself.<br /> <br /> In particular, it is useful to read these course materials and exercises: [http://lambda.ee/w/images/8/81/Chapter08-abbrev.pdf logic AIMA book], [http://lambda.ee/w/images/a/a6/Chapter07-abbrev2.pdf wumpus world in AIMA book], [http://lambda.ee/w/images/8/80/Chapter13-abbrev2.pdf uncertainty in AIMA book], [http://lambda.ee/w/images/5/5b/Chapter1415-abbrev.pdf probability models in AIMA book],<br /> [http://lambda.ee/wiki/Iti0210lab8 prolog lab], [http://lambda.ee/wiki/Iti0210lab9 bayes lab]<br /> <br /> == Focus ==<br /> <br /> The main focus of the course is on [https://en.wikipedia.org/wiki/Knowledge_representation knowledge representation and reasoning] (KR), on a spectrum from simple to very complex: representing and using knowledge in databases, sentences in natural language and commonsense knowledge. <br /> <br /> The overall goal of the labwork to understand issues facing a task of building a natural language question answering system a la [https://en.wikipedia.org/wiki/Watson_(computer) Watson] and to see and experiment with existing parts of the solution.<br /> <br /> The course contains the following blocks:<br /> <br /> * Background and basics. Representing and using simple facts and rules.<br /> * Knowledge conveyed by natural sentences: both statistical methods like word2vec and logic-based methods.<br /> * General-knowledge databases: wikidata, wordnet, yago, conceptnet, nell, cyc.<br /> * Reasoners and question answering systems.<br /> * Context, time, location, events, causality.<br /> * Different kinds and dimensions of uncertain knowledge.<br /> * Indexes and search.<br /> <br /> Check out [http://www.lisperati.com/tellstuff/ this description] of the whole KR area.<br /> <br /> == Books to use ==<br /> <br /> * Recommended for NLP: Speech and Language Processing by Dan Jurafsky and James H. Martin. [https://github.com/rain1024/slp2-pdf/blob/master/complete-book-pdf/slp2.pdf here is the 2nd edition] and [https://web.stanford.edu/~jurafsky/slp3/ here the web page for the draft 3rd edition]<br /> * Ca half of the course themes are covered in [http://www.sciencedirect.com/science/book/9781558609327 this book] with freely accessible pdf-s.<br /> * The freely accesible pdf of the [http://ii.fmph.uniba.sk/kri/KRhandbook.pdf handbook if knowledge representation] gives a detailed and thorough coverage of the subject; far more than necessary for the course.<br /> <br /> <br /> Observe that a noticeable part of the course contents are not covered by these books: use the course materials and links to papers, standards and tutorials provided.<br /> <br /> == Practical work ==<br /> <br /> There are three labs. They are all steps in a single project: build a simple natural-language question-answering system.<br /> <br /> The labs have to be presented to the course teachers and all students present at labwork time.<br /> <br /> The labs can be prepared alone or by teams of two people. The first lab task will be given on 6. February.<br /> <br /> First, read the explanation of the [[KR 2019 homework: overall task|overall task]]. The following labs are steps on the path to the end goal.<br /> <br /> === NB! You have to register for the lab like this: ===<br /> <br /> First, go to<br /> <br /> https://ained.ttu.ee/<br /> <br /> Search for: ITI8700<br /> <br /> Log in with the "UNI ID".<br /> <br /> Enroll on the course page found: Click "enrol me" button.<br /> <br /> Second, go to http://gitlab.cs.ttu.ee and log in<br /> <br /> Click on "Create a project"<br /> <br /> Name of the project must be exactly that: iti8700-2019<br /> <br /> Visibility level: private<br /> <br /> Upload you code, presentation, examples etc you use for the<br /> presentation.<br /> <br /> <br /> <br /> <br /> <br /> === First lab ===<br /> <br /> <br /> Deadline: <b>14. March</b> (after this there will be a penalty).<br /> <br /> If you manage to fullfill the task earlier, start doing the second lab.<br /> <br /> Please read about [[KR 2019 homework part 1: overall task|the task for the first lab]]<br /> <br /> === Second lab ===<br /> <br /> Deadline <b>15. May</b>. You have to give a presentation latest at 15. May labtime (better at 8. May).<br /> <br /> The task in the second lab is actually answering simple questions from the input text, using a small set of rules you write and a reasoner.<br /> <br /> Please start by experimenting with a real prover: look and run [[Reasoner examples with gkc]]<br /> <br /> We use the [https://github.com/tammet/gkc gkc] prover for examples (instructions are in the Reasoner examples ... page above) but you could also try out an old classic prover [https://www.mcs.anl.gov/research/projects/AR/otter/ Otter] which actually works OK on windows even though it is really old.<br /> <br /> Make sure to get the latest release of gkc from [https://github.com/tammet/gkc/releases gkc releases] (currently delta) and either download a binary or compile using instructions on the [https://github.com/tammet/gkc gkc github page]. The basic instructions are also on the [https://github.com/tammet/gkc gkc github page]. Also, see the few examples in the Examples folder.<br /> <br /> You can use a trivial NLP-to-reasoner parser in python [[nlp.py]] as a starting point.<br /> <br /> Have a look at the [http://lambda.ee/wiki/Teadmiste_formaliseerimine#Lecture_12:_seminar-kind_of_work_on_the_second_lab Lecture 12 materials].<br /> <br /> Interesting large question-answering datasets and challenge problems:<br /> [https://rajpurkar.github.io/SQuAD-explorer/ Stanford SQUAD], [http://data.allenai.org/arc/ Allen institute arc2 challenge], [https://ai.google.com/research/NaturalQuestions Google natural questions], [https://openinnovationgateway.com/ai-nlp-challenge/challenge.php Fujitsu NLP challenge], [http://jmcauley.ucsd.edu/data/amazon/qa/ amazon qa dataset]<br /> <br /> <br /> <br /> <br /> <br /> === Third lab ===<br /> <br /> Deadline <b>15. May</b>. You have to give a presentation latest at 15. May labtime (better at 8. May).<br /> <br /> The third lab is <b>optional</b> and will simply give as many points as lab 1 or 2 towards the final result and the grade: practical work 60% and exam 40%.<br /> <br /> The task in the third lab is the open-world scenario of answering questions based on wikipedia and using large downloaded rule sets a la yago, wordnet etc with a reasoner. <br /> <br /> It is a plus to be able to give uncertain answers (likely, unlikely) and handle fuzzy properties like boy, girl, grownup, rich etc.<br /> <br /> Have a look at the [http://lambda.ee/wiki/Teadmiste_formaliseerimine#Lecture_13:_seminar-kind_of_work_on_the_third_lab Lecture 13 materials]<br /> <br /> <br /> <br /> == Lecture block 1: basics and representing simple facts. ==<br /> <br /> <br /> === Lectures 1 and 2: Overview of the course. Background and basics: SQL, logic, NLP ===<br /> <br /> Lecture materials:<br /> * [[Media:Kr_lect_1b.ppt| intro, sql, logic, NLP]]<br /> <br /> <br /> === Lecture 2: RDF and RDFS (and OWL) ===<br /> <br /> Lecture materials:<br /> * [[Media:Kr_lect_2b.ppt| schemaless databases, RDF and RDFS]]<br /> * [[Nosql links and notes]]<br /> <br /> == Lecture block 2: capturing meaning in natural language ==<br /> <br /> === Lecture 3: Intro to homework and NLP ===<br /> <br /> Lecture materials:<br /> * [[Media:Kr_lect_3a.pdf| nlp and watson]]<br /> * Jurafsky and Martin book: [https://web.stanford.edu/~jurafsky/slp3/slides/convagents1.pdf conversational agents]<br /> * start reading the Jurafsky and Martin book (see links above)<br /> <br /> === Lecture 4: vector representation of words ===<br /> <br /> This lecture will be given by Priit Järv.<br /> <br /> Lecture materials:<br /> * First, listen to [http://lineardigressions.com/episodes/2017/4/23/feature-processing-for-text-analytics text vectorization episode] from the [http://lineardigressions.com/ linear digressions] podcast.<br /> * Second, listen to [http://lineardigressions.com/episodes/2018/12/30/re-release-word2vec word2vec episode] from the [http://lineardigressions.com/ linear digressions] podcast.<br /> * Read the beginning (until you get tired ...) of the [https://web.stanford.edu/~jurafsky/slp3/6.pdf Vector Semantics chapter] from the Jurafsky & Martin book.<br /> Useful additional materials from (roughly) easier to more complex:<br /> * [https://medium.com/@jayeshbahire/introduction-to-word-vectors-ea1d4e4b84bf word vectors]: a short intro<br /> * [https://web.stanford.edu/~jurafsky/slp3/slides/vector1.pptx vector semantics part I] presentation from the Jurafsky & Martin book for for chapter 6.<br /> * [https://towardsdatascience.com/art-of-vector-representation-of-words-5e85c59fee5 vector representation]: a short tutorial with code examples.<br /> <br /> * [https://www.tensorflow.org/tutorials/representation/word2vec word2vec tutorial] from [https://www.tensorflow.org/ tensorflow]<br /> * [https://web.stanford.edu/~jurafsky/slp3/slides/vector2.pptx vector semantics part II] presentation from the Jurafsky & Martin book for chapter 6.<br /> <br /> Probabilistic models:<br /> * [https://www.youtube.com/watch?v=u7l5hhmdc0M LDA explanation by Jordan Boyd-Graber]<br /> <br /> Also interesting to read:<br /> * [https://drive.google.com/file/d/0B8i61jl8OE3XdHRCSkV1VFNqTWc/edit Aetherial Symbols] by the ultimate machine learning guru aka "godfather of machine learning" [https://en.wikipedia.org/wiki/Geoffrey_Hinton Geoffrey Hinton]<br /> *[https://yurichev.com/blog/markov/ Autocomplete using Markov chains]<br /> <br /> == Lecture block 3 and start of 4: large common-sense knowledge bases and reasoning ==<br /> <br /> === Lecture 5: First look into main large knowledge bases ===<br /> <br /> We will have a look at the goals, main content and differences between:<br /> <br /> * [https://wordnet.princeton.edu/ wordnet]<br /> * [https://wiki.dbpedia.org/ dbpedia]<br /> * [https://www.mpi-inf.mpg.de/departments/databases-and-information-systems/research/yago-naga/yago/ yago]<br /> * [https://en.wikipedia.org/wiki/BabelNet babelnet]<br /> * [http://conceptnet5.media.mit.edu/ conceptnet]<br /> * nell in [https://en.wikipedia.org/wiki/Never-Ending_Language_Learning wiki] and [https://cacm.acm.org/magazines/2018/5/227193-never-ending-learning/fulltext a good paper]. The [http://rtw.ml.cmu.edu/ nell website] is currently down.<br /> * [https://en.wikipedia.org/wiki/Cyc cyc]<br /> <br /> === Lecture 6: Big annotation systems and intro to rule reasoners ===<br /> <br /> First, big annotation systems:<br /> * [https://developers.google.com/search/docs/guides/intro-structured-data Google structured data] on the webpage.<br /> * [https://developers.facebook.com/docs/sharing/webmasters/ Facebook Open Graph markup] on the webpage.<br /> <br /> In connection, see also:<br /> * [http://schema.org schema.org]: property markup vocabulary suggested by Google, Microsoft and others.<br /> * [https://json-ld.org/ json-ld]: currently most popular rdf syntax (in json), also recommended by Google.<br /> <br /> Second, intro to rule reasoners:<br /> * [[Media:Kr_lect_5a.pptx| Lecture material as ppt]] or as [[Media:Kr_lect_5a.pdf| pdf]]<br /> <br /> Additionally you may want to look at:<br /> * [http://lambda.ee/w/images/0/06/Geoffreasoningnotes.pdf automated reasoning tutorial by Geoff]<br /> <br /> === Lecture 7: rule reasoners ===<br /> <br /> Please have a look and run the same experiments yourself with the gkc prover:<br /> <br /> [[Reasoner examples with gkc]]<br /> <br /> We use the [https://github.com/tammet/gkc gkc] prover for examples (instructions are in the Reasoner examples ... page above) but you could also try out an old classic prover [https://www.mcs.anl.gov/research/projects/AR/otter/ Otter] which actually works OK on windows even though it is really old.<br /> <br /> <br /> <br /> <br /> <br /> === Lecture 8: starting to parse natural language for interaction with rule reasoners ===<br /> <br /> We will start building a tiny question-answering tool, translating natural language to rule reasoner input.<br /> <br /> Our first, very simple goal: translate the text "John is a father of Pete. Pete is a father of Mark. Who is the father of Pete?" to <br /> a rule reasoner input <br /> <pre><br /> father(john,pete).<br /> father(pete,mark).<br /> -father(X,pete) | ans(X).<br /> </pre><br /> then run the reasoner, fetch the ans(john) and give an answer "John is". <br /> <br /> You can use a trivial NLP-to-reasoner parser in python [[nlp.py]] as a starting point.<br /> <br /> We will then proceed to gradually make questions and sentences more complex and add rules to be able to answer them.<br /> <br /> In the final part of the course we will start looking at uncertain reasoning: right now we will only try to do simple strict reasoning.<br /> <br /> We will have a look at potential tools for semantic parsing, from simpler to more complex:<br /> * [https://github.com/Kitt-AI/parsetron parsetron]: a really simple and very limited tool to build your own parser.<br /> * [https://github.com/wcmac/sippycup sippycup]: has a detailed [https://nbviewer.jupyter.org/github/wcmac/sippycup/tree/master/ tutorial/course] attached, check the [https://nbviewer.jupyter.org/github/wcmac/sippycup/blob/master/sippycup-unit-0.ipynb first lab] <br /> * [https://github.com/mynlp/ccg2lambda ccg2lambda]: parses to classical logic strings, which are embedded in XML output.<br /> <br /> Other interesting things to look at:<br /> * [https://framenet.icsi.berkeley.edu/fndrupal/ Framenet]: map meaning to form in English through the theory of Frame Semantics.<br /> * [https://github.com/google/sling Sling]: Google project for getting facts from Wikipedia using Frame Semantics.<br /> * [https://yale-lily.github.io/spider Spider]: Yale Semantic Parsing and Text-to-SQL Challenge.<br /> * [https://github.com/salesforce/WikiSQL WikiSQL]: A Salesforce crowd-sourced dataset for developing NLP interfaces for relational databases.<br /> * [https://github.com/allenai/allennlp Allennlp] from the [https://allenai.org/ Allen Institute for AI]. See also their [https://github.com/allenai/acl2018-semantic-parsing-tutorial semantic parsing tutorial].<br /> <br /> == Lecture block 5: context and uncertainty ==<br /> <br /> <br /> === Lecture 9: intro to representing context and uncertainty ===<br /> <br /> Blocks world. <br /> <br /> * [http://www.ee.pdx.edu/~mperkows/May4/005.Ch11-planning.ppt Old intro: planning and blocks world]<br /> <br /> Some axiom sets:<br /> * [[PLA001-0.ax]]<br /> * [[PLA001-1.ax]]<br /> <br /> And some planning problems to solve from these axiom sets:<br /> <br /> * [[PLA004-1.p]]<br /> * [[PLA005-1.p]]<br /> * [[PLA019-1.p]]<br /> <br /> Also:<br /> <br /> [[blocks world axiom/query examples from lecture|Alternative axiomatization]] <br /> <br /> Next, starting uncertainty.<br /> <br /> We will first look at [http://en.wikipedia.org/wiki/Default_logic default logic].<br /> <br /> === Lecture 10: various classical ideas for discrete reasoning with uncertainty ===<br /> <br /> We will look at some discrete ways:<br /> <br /> * [http://en.wikipedia.org/wiki/Default_logic default logic]: refresher from previous lecture.<br /> * [https://en.wikipedia.org/wiki/Circumscription_(logic) circumscription] <br /> * Autoepistemic logic: long tutorial [[Media:Ijcai93.pdf|Ijcai93.pdf]]<br /> <br /> And start with numeric:<br /> <br /> * [[Media:Uncertain_prob_fuzzy.ppt‎|Uncertain_prob_fuzzy.ppt‎]] Intro to Bayesian and fuzzy reasoning.<br /> * [[Notes about numerical confidence, popularity, etc measures]]<br /> <br /> === Lecture 11: various classical ideas for numeric reasoning with uncertainty ===<br /> <br /> Classical numeric ways to reason about uncertainty:<br /> <br /> * [[Media:Uncertain_prob_fuzzy.ppt‎|Uncertain_prob_fuzzy.ppt‎]] Intro to both Bayesian and Fuzzy reasoning.<br /> <br /> Some newer approaches:<br /> <br /> * [[Media:Vienna_tanel_4.pdf|Vienna_tanel_4.pdf]] Additional examples and combining.<br /> * [https://dtai.cs.kuleuven.be/problog/ Problog] start with a [https://dtai.cs.kuleuven.be/problog/tutorial.html tutorial]<br /> <br /> === Lecture 12: seminar-kind of work on the second lab ===<br /> <br /> Background to start with:<br /> * Take and unpack [https://github.com/tammet/gkc/releases/download/v0.1-gamma/gkc.zip gkc.zip] from [https://github.com/tammet/gkc/releases/tag/v0.1-gamma gkc release gamma]<br /> <br /> * Start with the example file p1.txt<br /> <pre><br /> father(john,pete).<br /> <br /> % either the first or a second or both facts are true:<br /> father(pete,mark).<br /> <br /> % equivalent to (father(X,Y) & father(Y,Z)) => grandfather(X,Z).<br /> -father(X,Y) | -father(Y,Z) | grandfather(X,Z).<br /> <br /> -grandfather(john,X) | ans(X).<br /> </pre><br /> <br /> * Run <pre>gkc p1.txt</pre><br /> <br /> * Start with the trivial NLP-to-reasoner parser in python [[nlp.py]] .<br /> <br /> * Our initial goal is to process the text "John is a father of Andrew. Andrew is a man. Who is the father of Andrew?"<br /> <br /> <br /> * [[Reasoner example created by nlp.py]]<br /> * [[Reasoner example with some rules for relatives]]<br /> <br /> And the examples with various ways to encode normal/abnormal cases (not really needed for lab):<br /> * [[Reasoner example for being likely married]]: works out ok<br /> * [[Reasoner example for birds]]: does not work out well<br /> * [[Reasoner another example for birds]]: does not work out well<br /> <br /> Syntax and command line:<br /> * gkc command line: [https://github.com/tammet/gkc gkc readme in github]: see also Examples folder for a few trivial examples.<br /> * For simple clausal syntax see [https://www.mcs.anl.gov/research/projects/AR/otter/otter33.pdf otter manual] clause syntax.<br /> * For complex syntax options see [http://tptp.cs.miami.edu/TPTP/TR/TPTPTR.shtml TPTP manual]: seems to be sometimes down.<br /> <br /> === Lecture 13: seminar-kind of work on the third lab ===<br /> <br /> Ways to get external data in addition to our own rules and data:<br /> <br /> * Raw data (hard to use): wikipedia, wikidata<br /> * Processed raw data (a bit easier): dbpedia (processed wikipedia/wikidata)<br /> * Well-formed and connected options (use one of these):<br /> ** [https://wordnet.princeton.edu/ wordnet]: exists in several versions, including an external [http://tptp.cs.miami.edu TPTP] version usable by the gkc reasoner, direct link to [http://tptp.cs.miami.edu/cgi-bin/SeeTPTP?Category=Axioms&File=NLP001+0.ax TPTP axiom NLP001+0.ax]<br /> ** [https://www.mpi-inf.mpg.de/departments/databases-and-information-systems/research/yago-naga/yago/ yago], connecting wikipedia with wordnet and geonames. Look at the [https://www.mpi-inf.mpg.de/departments/databases-and-information-systems/research/yago-naga/yago/downloads/ downloads section] SIMPLETAX and download [http://resources.mpi-inf.mpg.de/yago-naga/yago3.1/yagoSimpleTypes.tsv.7z yagoSimpleTypes] and [http://resources.mpi-inf.mpg.de/yago-naga/yago3.1/yagoSimpleTaxonomy.tsv.7z yagoSimpleTaxonomy]. You will need to write a program to filter and convert the yago datasets to the prover-understandable form. See [[some ideas for converting yago]].<br /> <br /> See [[how to run gkc on the Wordnet TPTP version]]<br /> <br /> <br /> [[Category:Kursus]]</div>

Tammet