Teadmiste formaliseerimine

teadmised

Name: Knowlege representation Code: ITI8700 Link: http://lambda.ee/Teadmiste_formaliseerimine Lecturer: Tanel Tammet tanel.tammet@ttu.ee ICT-426 Practice sessions: Riina Maigre Archives of previous years: 2025.

Eksamitulemused

Eksamitulemused siin:

• Esimene (27 mai) teadmiste eksam 2026
• Teine (1 juuni) teadmiste eksam 2026
• Kolmas (8 juuni) teadmiste eksam 2026

Eksamikonsultatsioon

(a) väike salvestuse teamsi, teemaks siis eksamikorraldus, ülesanded jne:salvestus siin, asub kursuse teamsi-salvestuste kataloogis.

(b) Konsultatsioonide lingid siin:

• eksamikonsultatsioon 20. mai kell 10:30-11:00. (enne seda, kell 10:00 hajussüsteemide konsultatsioon, tolle kursuse lehel oleval lingil).
• eksamikonsultatsioon 25. mai kell 10:30-11:00. (enne seda, kell 10:00 hajussüsteemide konsultatsioon, tolle kursuse lehel oleval lingil).

Eksamikorraldus ja materjalid õppimiseks

Eksamiks on kokku aega 2.5 tundi, aga enamik peaks saama hakkama ca 1.5 tunniga.

Eksam on kirjalik, samas ruumis/ajal toimub ka hajussüsteemide kirjalik eksam.

Allolevad punase E1-ga on eksami jaoks kõige olulisemad materjalid, E2 veidi vähem olulised. NB! Võrreldes varasemaga on lisaks markeeritud E2-ga need varasemalt olemas, aga markeerimata materjalid:

• Lecture material part 1 E2
• Lecture material part 2 E2
• Lecture material E2
• Lecture presentation E2

Ekamil tuleb neli eraldi hinnatavat osa: kolm teemaküsimust ja üks osa paari lühiküsimusega. Spetsiifiliste süntaksite (RDF, loogikavarandid jne) osas: ühtegi täpset süntaksit/võtmesõnu ei pea teadma, aga esitus peab olema õppejõule sisuliselt arusaadav. Millised on teemaküsimused ja millised on lühiküsimused:

SQL ja loogika: esitan paar tabelit ja küsin mõnda taolist küsimust:

• nende sisu võimalikult täpset esitust esimest järku loogikas või RDFs
• mõne konkreetse where- või join-päringu kirjutamist esimest järku loogikas

JSON-LD: annan JSON-LD jupi ja küsin selle tähendust ehk teisendust RDF kujule kolmikutena.

Reeglite esitamine: pane rdf ja/või rdfs kujule ja/või predikaatloogikale baseerudes kirja etteantud konkreetsed inimkeelsed väited.

Reeglite rakendamine klassikalises teoreemitõestamises: oskus teha käsitsi läbi väike resolutsioonimeetodi rakendamise näide, kui disjunktid ette antud.

Sõnavektorid ja N-grammid ja keelemudelite üldprintsiibid: arusaamine, oskus esitada väike näide töötamise kohta, võimalikud kasutus- või loomisstsenaariumd

Ebakindlate teadmiste esitamise põhimeetodid: pidevad (erinevat sorti tõenäosuste-tüüpi esitus) ja diskreetsed (eranditega süsteemid, eeskätt default logic). Siin võin tuua väikese näite ja küsida lahendust või palun ise esitada mõne pisinäite sobival kujul. Konkreetseid süntakseid teadma ei pea, piisab õppejõule arusaadavaks tegemisest.

Lühiküsimused (siit valitakse iga eksami jaoks paar):

Seleta paari lausega, mis asi on, vahel küsin ka näidet: freimiprobleem, open/closed world, blocks world, RAG, erinevad tõenäosuste interpretatsioonid, RDFa; mis asi on ja mida sealt umbes leida võib ja mida pigem mitte: schema.org, wordnet, conceptnet, yago, cyc; mida sa mäletad "semantiline parsimine" ja "visual reasoning" teemade kohta.

Eksamipäevad ja ajad

Siin on täpsustunud esimese kahe eksami ajad:

• 27 mai, kolmapäev, kell 11:00 (koos hajussüsteemide eksamiga). Auditoorium NRG-226.
• 1 juuni, esmaspäev: eraldi eksamiajad hajussüsteemide (11:00) ja teadmiste esitamise (14:00) jaoks. Auditoorium SOC-211a.
• 8 juuni, esmaspäev: kell 11:00 (koos hajussüsteemide eksamiga). Auditoorium ICT-315.

Time, place, result

Lectures (Tanel): Mondays 12:00-13:30 room SOC 414
Practice sessions (Riina): Tuesdays 08:00-09:30 room MEK-122 MEK-109.

The lectures and sessions will be on site (physical room) plus teams. Later we may decide to do some lectures/practice sessions on teams only: this will be advertised on the page here early. Beware: recording may not always succeed.

Please join the teams group of the course. The code for joining is zb3z457.

Weekly teams link for the lecture.

Weekly teams link for the practice session.

Grading

Practical work will give 50%, exam 50% of points, and weekly tasks additional extra up to 10%, underlying the final grade. The exam will consist of (a) several questions asking for the explanation of important concepts along with examples (b) some small excercises.

The materials below marked with the red E1 are the main materials for the exam (10 of these on the page), while E2 materials (20 on the page) will also come up, but less and/or in a simpler form.

To finish the course successfully, you need to

• successfully pass three practical works
• receive at least 1/3 of the max points at exam
• get practice work + weekly tasks + exam at least 50% points altogehter.

Contents overview: what the course is about

The theme of the course: from SQL to natural language.

Hence the main focus of the course is on hybrid methods for knowledge representation and reasoning: symbolic AI, machine learning / neural methods and their neurosymbolic combinations. We look at this spectrum from simple (representing and using knowledge in databases) to complex (meaning of sentences in natural language and commonsense knowledge) tasks. A closely related subject is commonsense reasoning.

The course contains the following blocks:

• Knowledge in SQL, RDF and JSON
• Handling simple facts and rules in symbolic AI and logic
• General-knowledge databases
• Natural language and question answering
• Uncertain knowledge and hybrid systems

Weekly tasks

The small weekly tasks are in addition to the practical work: their goal is to practice with the stuff presented at the last lecture.

The first weekly task

Ülesanne on eksperimenteerida veidi logictools.org saidiga ja teha ise läbi loengus toodud näited SQL data ja päringute esitamisest loogikas. Tähtaeg 17. veebruar.

Konkreetselt tee järgmist:

• Eksperimenteeri veidi logictools.org "simple examples" valikuboksi näidetega basic ... equality, loe nende selgitusi ja püüa aru saada tõestustest.
• Loe seejuures predikaatloogika väikest seletust about lehel kuni lausearvutuseni (seda pole enam otse vaja lugeda).
• Kirjuta ise valmis näide vastavalt kursuse materjalile Background: relations of sql and logic lk pealkirjaga "SQL join as logical rule", kus on toodud ühe client tabeliga ja client ning cars tabelitega näide sql päringute teisendamisest loogikasse (üks lihtne select ja üks join):

client table: id name balance
1 john 100 
2 pete -200

cars table: id model owner
1 ford 1
2 opel 2
3 saab 2

select client.name, client.balance
from client where balance<0

select client.name, cars.model from client, cars
where client.id=cars.owner

Tee need näited reaalselt valmis nii, et logictools.org sööb nad sisse ja annab ootuspärase vastuse. Seejuures on sul vaja kasutada $ans predikaati, nagu "answers" näites toodud. Kui tahad mitut vastust (ei ole kohustuslik), siis vaata ka "multiple answers" näidet.

• Saada näite-sisendid ja genereeritud tõestused õppejõule emailiga (riina.maigre at taltech.ee ja cc tanel.tammet at taltech.ee). Kui jääd nendega hätta, siis kirjuta, mida proovisid, too oma näitetekst ja selgita, mille juures hätta jäid (see on ka arvestatav tulemus): kirjuta aadressile riina.maigre at taltech.ee või tanel.tammet at taltech.ee ja pane pealkirja sisse mh sõna "formaliseerimine", et kirja kergesti üles leiaks.

• Kasuta kindlasti oma taltech.ee aadressi, sest muidu võib kiri lõpetada rämpspostis!

Soovitused:

• Negatiivset arvu sa logictoolsi vaikimisi süntaksiga otse sisestada ei saa: -200 interpreteeritakse kui loogilise eituse rakendamist 200-le. Kaks head võimalust negatiivse arvu sisestamiseks on kas kirjutada 0-200 või $difference(0,200).
• Aritmeetika jaoks vaata help teksti Arithmetic peatükki.
• Väiksem-tingimuse jaoks vaata complex examples valikuboksis "arithmetic" näidet.

The second weekly task

Selle ülesande mõte on veidi praktiseerida JSON-LD esitusega. Sinu konkreetne ülesanne on võtta oma esimene nädalaülesanne ja esitada needsamad andmed (ja soovitavalt ka reeglid) JSON-LD kujul.

Vaata alustuseks json-ld näiteid logictoolsi json-tabi alt. NB! Enamus näiteid "example" valikuboksist ei ole konventsionaalne json-ld, aga mitmed on.

Practical work

There are three labs.

The labs have to be presented (as a demo + small overview of the software and principles) to the course teachers and all students present at labwork time, either in the auditorium or over MS teams. The labs can be prepared alone or by teams of two or three people.

Each lab is evaluated independently and gives the same number of points. NB! When you do not meet the lab deadline, you will be awarded half of the points for this lab.

First lab

The 2026 task for the first lab is about encoding nontrivial information in various ways (standard logic, rdf, json-ld) and writing simple logical rules for answering questions.

Deadline: 10. March.

Second lab

The 2026 task for the second lab is about using a VLM (visual language model) to get a structured description from an image which would fit with the pre-existing simple world model.

Deadline: 7. April

Third lab: questions in NLP

This is the task for 2026:

Experiment with neurosymbolic reasoning: LLM as a parser plus a symbolic reasoner for images: 2026

Deadline: 12. May

Books to use

Main materials:

• Get the basic background about classical automated reasoning from the book chapter pages 297-345 by Tanel Tammet and optionally more details from the set of course notes from Geoff Sutcliffe.
• The book for natural language (NLP): Speech and Language Processing by Dan Jurafsky and James H. Martin. the web page for the draft 3rd edition. Recommendation: read from the beginning up to an including chapter 6: Vector Semantics and Embeddings, then read chapter 6 about extracting information and the appendix chapter Logical Representations of Sentence Meaning.

Other useful books and tools for symbolic K&R:

• logictools.org for experimenting with classical automated reasoning
• Symbolic reasoning themes are covered in this book with (hopefully) freely accessible pdf-s via Taltech. See also slides for the book.
• Similarly, the freely accessible pdf of the handbook of knowledge representation gives a detailed and thorough coverage of the subject; far more than necessary for the course.
• You may want to get a small intro to Prolog by reading the first 100 pages of a classic Art of Prolog book. It is not strictly necessary for the course (we will not be using Prolog) but it surely helps. Besides, the book is interesting, quite easy to read and works out a lot of interesting examples.

Other useful material for neural NLP:

• Andrej Karpathy nanoGPT and neural networks zero to hero course

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.

Blocks and lectures: the main content of the course

The details and materials of lectures in the future are from 2025 and will be revised/extended. The materials of the current or passed lectures are up to date.

Block 1: knowledge in SQL, RDF and JSON

Lecture: Intro and background: 2. February

Intro presentation:

teadmised_intro2026.pdf

Recommended non-obligatory listening: Francois Chollet, Yann LeCun, Gary Marcus, Demis Hassabis, Ilya Sutskever, Ben Goertzel ,

Lecture: Nosql, json and rdf in databases: 9. February

Lecture materials:

• We cover Background: relations of sql and logic E1

• and start with schemaless databases, RDF and RDFS E1 (also important for practice work)

See also

• Json in Postgresql: json datatype and functions for json

The weekly homework will be about trying out small things in logic: please try out logictools from "basics" to "equality" in simple examples selectbox. Details will appear after the lecture.

Block 2: handling simple facts and rules in symbolic AI and logic

Lecture: simple rules in rdfs, logic and owl + reasoner: 16. February

• Main lecture material: schemaless databases, RDF and RDFS

Rdf example developed during lectureE2

Have a brief a look at this, no need to read thoroughly:

• Official w3c rdf and rdfs primer E2

The main materials covered in the lecture:

• json-ld (check out playground examples) E1 currently most popular triple representation language on top of json. see also wikipedia and w3c standard: for the latter, read the basic concepts chapter E2.
• rdfa E2 triple markup language suited for html, see also wikipedia
• schema.org (E2 you can surf on the schema.org site and understand/find answers there)) we looked at in the last lecture: property markup vocabulary suggested by Google, Microsoft and others.

And then have brief look (skim through) at:

• Official w3c owl primer but it is better to look at this intro presentation and continue with the owl rules presentation
• sparql tutorial and wikipedia take on sparql.

Certainly read:

• Hea jutt Whatever Happened to the Semantic Web?

We will also start at looking and understanding the examples in logictools.org

More resources:

• sparql implementations in wikipedia
• various rdf tools listed at the bottom of the page

What the reasoner does:

We will consider the reasoning part of K&R to understand what provers do:

• rules, logic, provers E1
• logictools system running gkc reasoner in the browser: try out the "simple" and "complex" examples, read explanatory texts and the manual about predicate logic
• Have a look at the gkc github repo, including precompiled binaries for command line in the release 0.6.0 and the tutorial Examples/Readme.md.
• Travel examples from lecture 2021

You may want to have a look at the additional materials - from easier to harder - from the

• book chapter pages 297-345 and optional extras, 346-382; by Tanel Tammet.
• the set of course notes from Geoff Sutcliffe,
• the implementation presentation from Stephan Schulz
• 5-lecture prover course from the authors of Vampire
• and the readable-and-hackable PyRes reasoner for educational use.

Lecture: DPP ontology development. 2. March

Loengu annab Riina. Digital Product Passport (DPP) kui näide praktilisest ontoloogia/sõnastiku arendamisest CIRPASS projektis.

Hetkeseisu algusleht.

CIRPASS-2

Ühikute ontoloogia

SHACL

UNTP DPP

Lecture: Next homework and looking into main large knowledge bases: 9. March

We will first present the next homework.

Then we will continue with existing large knowledge bases.

Slides:lecture slides E2, a tool to explore Wordnet taxonomies can be found here: https://github.com/martinve/wntool

You can search/investigate what kind of data is available and you can find out actual data from these databases with some work, by surfing on the web pages of the systems.

About knowlege base creation using LLMs: have a look at this paper.

We will have a look at the goals, main content and differences between:

• wordnet see tptp axioms
• dbpedia, see also classes and dbpedia wiki
• wikidata
• yago old page and new page
• babelnet described in wikipedia
• conceptnet described in wikipedia
• nell website (sometimes down) and in wiki and a good paper.
• framenet described in wikipedia, see example
• cyc but just read wikipedia and see whitepaper also tptp small random selection of axioms
• sumo and adimen-sumo, see sumo in wikipedia and tptp axioms
• tptp: a large set of axioms and problems in logic usable for automated reasoners
• schema.org: property markup vocabulary suggested by Google, Microsoft and others.
• Wolfram alpha stuff

Block 3: natural language and question answering

Lecture: Intro to NLP, n-grams, word vectors: 16 March

• These three wikipages give useful introductory details:Natural language processing, Knowledge extraction, Natural language understanding.
• Tanel Alumäe keelemudelitesse sissejuhatavad slaidid E1 ja loengusalvestus
• Väga hea õpik: 3. variant.
  • Vektorsemantikast arusaamiseks loe 3. variandi ptk 3. N-gram Language Models E2 ja siis chapter 5: Embeddings E2. Vaata sisukorrast ka slaide nende kohta.
• Google ngrams ja reaalne data.
• GloVE: a relatively simple vector representation
• Kui tahad katsetada toore Wikipedia dataga, siis saab loengus vaadatud töödeldud variante siit (tarballid sisaldavad datat, selgitavat README-d ja tarkvara selle ise ehitamiseks): a compacted pure-text version of full wikipedia, A lemmatized version of wikipedia texts, Several co-occurrence matrices and lists of top-co-occurring words for wikipedia
• Loengus tehtud katse Glove vektoritega selle jutu järgi.

Lecture: large language models (LLM) 23 March

• Lecture material part 1 E2
• Lecture material part 2 E2

Eesmärk: saada pinnaliselt aru LLM ja transformerite (NLP masinõppe põhiasjad) peamistest ideedest. Detailid on tegelikult keerulised ja nendest arusaamine võtab hulga rohkem aega, kui meil kursuses on, sestap lepime vähesega. Samas, kui sul endal huvi, siis loomulikult on teretulnud detailsem iseõpe.

Enne järgmisse materjali süvenemist tutvu kindlasti vektorsemantikaga, mida seletasime 11 märtsi loengus.

Mida vaadata ja lugeda:

Eriti soovitav

• Andrej Karpathy progemisega-koos loeng: vaata see süvenemisega läbi! Vt ka Karpathy.

Paratamatult jääb enamus detaile loengus vähearusaadavaks (kui ise ei kuluta hulga aega järele katsetamiseks), aga sellegipoolest on see tõenäoliselt kõige parem loeng transformeritest: mingi tunnetuse sellest, et mis toimub, saab.

Pikemaks lugemiseks:

• visualizing neural machine translation and then the illustrated transformer
• Jurafsky & Martin book: neural networks and the slides. Alternative: CMU Mitchell course: neural networks basics
• Jurafsky & Martin: chapter 9, transformers

Optsionaalselt üldisem arusaam klassikutelt:

• Jann LeCun: värske episood Lex Fridmani podcastist
• Wolframi pikk jutt lugemiseks koos närvivõrkudega üldisema taustaga
• Rodney Brooksi ennustused transformerite kohta

Optsionaalselt diipim arusaam:

• nanoGPT: [Karpathy reaalne näitekood GPT ise-ehitamiseks (otse eelmise loenguga seotud) : seda saab ilma suuremate pingutusteta käima linuxil ja macil, windowsil tekib aga suuremaid probleeme (proovisin ise linuxil järgi).
• Taustaks: Karpathy terve loenguseeria, kui sul tekib sügavam huvi
• Jurafsky & Martin ptk 10 Large Language Models kui tahad ise juurde lugeda.
• Good free (?) book: Build a Large Language Model (From Scratch)
• väga detailne ja mahukas videoloengutega deep learningu kursus, mh raamat.

Lecture: RAG usage pattern / context engineering for LLMs : 30 March

Lecture material: this is focused on RAG and written in 2024 and spring 2025: good tool use, good reasoning and more autonomy like claude code and openclaw are not covered in this overview.

"Context Engineering" became a popular phrase in 2025: essentially it means "fancy RAG", i.e. significant effort put towards creating a really good set of additional information (prompt additions: "context") for each user query/prompt. It can also mean approaches for storing such additional information on the fly in special files which are then used RAG-like as extra input in addition to user dialog.For example, claude code memory principles and memory folder structure, openclaw memory.

Useful reading in addition to pointers in the lecture material:

• application landscape overview
• RAG focus: LLM improvement for applications and a basic overview of customizing llms
• Old finetuning example: Finetuning llama 2
• RAG:
  • simplest RAG intro E1
  • more rag details: a good architecture picture, may skip the improvements part after that
  • encoder/decoder and Seq2seq as a generalization of word2vec and openai embedding model api
  • vector database E2 as a useful tool
  • different kinds of rag E2 based on paper

Block 4: hybrid systems and uncertain knowledge

We will focus on (a) world modeling and uncertainty in KR (open/closed world, frame problem, probabilities, exceptions) and (b) current research on building hybrid systems

Lecture: issues in formalizing commonsense reasoning: open/closed world, frame problem, uncertainties: 6. April

First, open/closed world, world states, frame problem

The standard example for these is a blocks world: some blocks are on the table and the robot arm can lift single blocks and put them on other blocks or on the table. The arm cannot lift several blocks or a tower of blocks. Then, given a position of blocks at initial situation, can the robot arm create a required new position of blocks? Can we get the steps required? For getting the steps we use the $ans predicate.

One hard issue arising is that how do we know that doing some action like lifting a block does not create side effects like stumbling existing towers, moving other blocks etc? This issue is called the frame problem and it has no easy solutions.

Importantly, the frame problem arises since normal first order logic has

• the open world assumption: if we do not have (and cannot derive) a positive fact like P(a) and neither a negative fact like -P(a), we assume we simply do not know which holds.

Prolog and databases operate under the

• closed world assumption E1 (must read): if we do not have (and cannot derive) a positive fact like P(a), we automatically assume that -P(a) must be true. For example, Prolog "negation" really means "cannot prove". One consequence of this assumption is that we cannot represent real negation and disjunction in an OK manner (Prolog does not contain these) and cannot satisfactorily speak about "true", "false" and "unknown". See also negation as failure in prolog

Please look at and read these materials about the frame problem:

• overview presentation: E1 must read.
• a classic AI philosophy paper about the frame problem: must read the first three pages (rest is optional).
• wikipedia about frame problem: E2 skim it through (no need to read all the details) to get a rough idea about different approaches to the frame problem. None of these are really good.

These two readings are optional:

• frame problem at the Stanford Encyclopedia of philosophy: not obligatory, but a good read to get a better understanding.
• another classic about frame problem: read this only if you became really really interested about the whole issue: it goes quite deep, although it is not especially technical.

You may also want to have a look at the algorithmic complexity and efficient solution algorithms, regardless of how we formalize the problem: see this article.

Next, importantly, you should experiment yourself with gkc and both of the following files. Please copy and read the files and understand the encoding. At the end of the files are several queries to experiment with: instructions are also there. The simplest way to experiment is to use https://logictools.org/ . There is already a blocks world example available.

In both of them the predicate holds(X,Y) is used to describe something X about the blocks world (like, one block is on another or robot holds some block) and the second argument Y describes the state we have arrived to via robot actions from the original state (like, picked up a block in the initial state and then did put the block down on top of another).

• Simple axiomatization of the blocks world: E2 please also read through the comments explaining the axiomatization and queries!
• More complex blocks world axiomatization taken from the TPTP problem set for first order logic, concretely by concatenanting and commenting axiom sets PLA001-0.ax, PLA001-1.ax and queries PLA004-1.p, PLA005-1.p, PLA019-1.p

To get a simple visualization of the blocks world, have a look at this tiny video.

Second, uncertainty

Loengu sissejuhatava osa presentatsioon: Reasoning_with_uncertainty.pdf E1.

Almost all the knowledge we have is uncertain: there are many exceptions to a rule or a fact/rule holds with some vague probability. Notice that, in contrast, typical databases in companies contain facts with very high certainty, but they do not contain rule or commonsense knowledge: the latter is built into brittle software code of systems using these databases.

We will consider two different ways to tackle uncertainty:

• Numeric methods: uncertainty is estimated with probability-like numbers of various kinds.
• Discrete methods: no numbers used. Instead, uncertainty is described as exceptions or beliefs of people.

NB! Both of these ways are hard to actually implement, neither have they been understood very well, despite an immense amount of research and papers. For example, typical machine learning methods operate with extremely vague probability-like numbers and do not attempt to put these numbers into a real properly-theoretical-and-correct probabilities framework.

The main material for reading about numeric methods is an overview-followed-in-depth by the lecturer Numeric uncertainty: (E2 up to and including the chapter 3, see also E2 tags inside)

• Read carefully up to and including the chapter 4 "Different ways to encode confidences in logic" and have a quick look at all the wikipedia links. In particular, Wiki on Bayesian inference and this
• Skim through the later parts and look only into these subchapters which you found interesting.

Next,

• Read the paper E2 about confidences. containing intro, algorithms and experiments. Then look at the web page with the confer reasoner and a wealth of examples.
• Read the paper E2 about exceptions (default logic). containing intro, alorithms and experiments. Then look at the web page with the gk reasoner and a wealth of examples.

Next, have a brief look at a good example system doing some specific kinds of probability reasoning:

• Problog start with a tutorial E2
• Try out some examples via browser from the tutorial

Lecture: research in neurosymbolic reasoning: 13 April

lecture slides

The lecture will be given by our PhD student Otto Mättas.

Lecture: semantic parsing with LLMs and with UDP: 20 April

The focus of the lecture is on semantic parsing.

For a quick intro we will use Martin's slides from 2024 semantic parsing E2.

The rest of the lecture will explain stuff in nlpsolver. See the nlpsolver presentation.

Useful background:

About important concepts in parsing:

• Part-Of-Speech (POS) and POS tagging
• Jurafsky&Martin: Constituency Parsing
• Jurafsky&Martin: Dependency Parsing

Some important toolkits and representation schemes:

• spacy, a small spacy UD parsing example
• stanza
• universal dependencies

Some draft parsing papers:

• concepts, based on UD
• pipeline, based on UD
• experiments with LLM usage

Lecture: we explore some recent papers in neurosymbolic reasoning: 27 April

A reading session: we will read and attempt to understand together some case studies.

• Google DeepMind paper Solving olympiad geometry without human demonstrations
• Google DeepMind paper Olympiad-level formal mathematical reasoning with reinforcement learning
• Proving Olympiad Inequalities by Synergizing LLMs and Symbolic Reasoning
• Concept-RuleNet: Grounded Multi-Agent Neurosymbolic Reasoning in Vision Language Models
• What’s Left?Concept Grounding with Logic-Enhanced Foundation Models

Lecture: visual reasoning: 4. May

The lecture will be given by our postdoc researcher Mohit Vaishnav.

Lecture presentation E2

Related preprints:

• Cognitive paradigms for evaluating vlms on visual reasoning task
• Symbolic Grounding Reveals Representational Bottlenecks in Abstract Visual Reasoning

Lecture: we continue exploring recent papers in neurosymbolic reasoning: 11 May

We look at some of the more interesting papers from each subarea (maybe not all the papers):

Logic focus:

• Neuro-Symbolic Concepts https://arxiv.org/pdf/2505.06191
• Sound and Complete Neurosymbolic Reasoning with LLM-Grounded Interpretations https://arxiv.org/pdf/2507.09751

Visual reasoning:

• MuSLR: Multimodal Symbolic Logical Reasoning https://arxiv.org/pdf/2509.25851
• Can VLMs Reason Robustly? A Neuro-Symbolic Investigation https://arxiv.org/pdf/2603.23867

Planning:

• LLMs as Planning Formalizers: A Survey for Leveraging Large Language Models to Construct Automated Planning Models https://arxiv.org/pdf/2503.18971
• Planning in the Dark: LLM-Symbolic Planning Pipeline without Experts https://arxiv.org/pdf/2409.15915 or newer https://ojs.aaai.org/index.php/AAAI/article/view/34855/37010
• Fast and Accurate Task Planning using Neuro-Symbolic Language Models and Multi-level Goal Decomposition https://arxiv.org/pdf/2409.19250
• See also PDDL https://fareskalaboud.github.io/LearnPDDL/ https://homepages.inf.ed.ac.uk/mfourman/tools/propplan/pddl.pdf

Text-to-SQL:

• SPIDER 2.0: EVALUATING LANGUAGE MODELS ON REAL-WORLD ENTERPRISE TEXT-TO-SQL WORKFLOWS https://arxiv.org/pdf/2411.07763
• CHASE-SQL: Multi-Path Reasoning and Preference Optimized Candidate Selection in Text-to-SQL https://arxiv.org/pdf/2410.01943