AI KR Learning Resource

This page intends to be a guide for learners as well as and curriculum development.

Please consider contributing as follows: 1. insert links to resource where missing, verify that the links provided are accessible 2. improve format/style the page 3. add additional resources 4. annotate your comments (did you use a resource? did you find it useful?) Thanks!

===========

Knowledge representation involves structuring information in a form that a computer can understand. A way to do this is by using ontologies or knowledge graphs, which allows for relationships and hierarchies within your data to be represented. (Source)

By the same token, it can be used to make the AI/ML black box explainable and human readable (Paola Di Maio, W3C AI KR Chair aka PDM)

Credit: PDM

Knowledge Representation (KR) is a broad field. In relation to Artificial Intelligence (AI) KR is concerned with how knowledge can be represented symbolically and manipulated in an automated way by reasoning programs. Here are some of its major subdomains:

Logic-Based Representation

This is arguably the most foundational subdomain. It uses formal logic to represent knowledge.

   Propositional Logic: Represents facts as propositions (statements that are either true or false).

   Predicate Logic (First-Order Logic): A more expressive logic that allows for variables, quantifiers, and relations between objects.

   Description Logics: A family of formal knowledge representation languages. They are more expressive than propositional logic but have better computational properties than first-order logic for many reasoning tasks. Widely used in a_n_ Semantic Web and ontologies.
 
 Modal Logics: Extend classical logic to deal with modalities like possibility, necessity, belief, and knowledge.

   Temporal Logics: Used to represent and reason about time-dependent information.

Ontologies and Semantic Networks

These focus on representing the relationships between concepts.

   Ontologies: Formal, explicit specifications of a shared conceptualization. They define concepts, properties, relationships, and axioms within a specific domain. OWL (Web Ontology Language) is a common standard.
   Semantic Networks: Graph-based structures that represent knowledge as a network of nodes (representing concepts or objects) and links (representing relationships between them).

Frame-Based and Object-Oriented Representations

These represent knowledge in terms of objects or frames with associated attributes (slots) and values.

   Frames: Data structures that represent stereotyped situations or objects. Slots in a frame can have default values, constraints, or attached procedures.
   Object-Oriented Representations: Similar to frames, but with stronger emphasis on encapsulation, inheritance, and methods (procedures associated with objects).

Rule-Based Systems (Production Systems)

These represent knowledge as a set of "IF-THEN" rules.

   An inference engine processes these rules to deduce new facts or make decisions. These are common in expert systems.

Probabilistic and Uncertainty Representations

These deal with knowledge that is uncertain or incomplete.

   Bayesian Networks: Probabilistic graphical models that represent conditional dependencies among a set of random variables.
   Dempster-Shafer Theory: A mathematical theory of evidence that can be used to combine evidence from different sources and arrive at a degree of belief.
   Fuzzy Logic: Deals with reasoning that is approximate rather than fixed and exact. It allows for degrees of truth.

Non-Monotonic Reasoning

This area addresses reasoning where the addition of new information can invalidate previous conclusions.

   Default Logic: Allows for reasoning with default assumptions that can be retracted if new evidence contradicts them.
   Circumscription: A formalization of the common-sense assumption that things are as expected unless otherwise specified.

Case-Based Reasoning (CBR)

This approach solves new problems by retrieving and adapting solutions to similar past problems (cases).

   Focuses on storing and indexing past experiences for reuse.

Multi-Agent Systems Knowledge Representation

Concerns how multiple autonomous agents represent their own knowledge, knowledge about other agents, and shared knowledge to coordinate and communicate effectively.

These subdomains often overlap, and many practical KR systems integrate techniques from multiple areas. The choice of representation often depends on the specific problem, the type of knowledge being represented, and the reasoning tasks required.

Read about AI KR The Knowledge Level (A. Newell) *link to paper

McCarthy, John, and Patrick J. Hayes. "Some Philosophical Problems from the Standpoint of Artificial Intelligence." Machine Intelligence, vol. 4, 1969, pp. 463–502.

   Schank, Roger C., and Robert P. Abelson. "Scripts, Plans, Goals, and Understanding: An Inquiry into Human Knowledge Structures." Artificial Intelligence, vol. 8, no. 3, 1977, pp. 241–273.

   Brachman, Ronald J. "On the Epistemological Status of Semantic Networks." Associative Networks: Representation and Use of Knowledge by Computers, Academic Press, 1979, pp. 3–50.

   Minsky, Marvin. "A Framework for Representing Knowledge." The Psychology of Computer Vision, edited by Patrick H. Winston, McGraw-Hill, 1975, pp. 211–277.

   Reiter, Raymond. "A Logic for Default Reasoning." Artificial Intelligence, vol. 13, no. 1–2, 1980, pp. 81–132.

   Davis, Randall. "Applications of Meta-Level Knowledge to the Construction, Maintenance, and Use of Large Knowledge Bases." Artificial Intelligence, vol. 18, no. 2, 1982, pp. 165–192.

   Lenat, Douglas B., and R. V. Guha. "Building Large Knowledge-Based Systems: Representation and Inference in the Cyc Project." Addison-Wesley, 1989.

   Brachman, Ronald J., Hector J. Levesque, and Raymond Reiter, editors. Readings in Knowledge Representation. Morgan Kaufmann, 1985

Study AI KR

to be linked

   Brachman, Ronald J., Hector J. Levesque, and Raymond Reiter, editors. Readings in Knowledge Representation. Morgan Kaufmann, 1985

   van Harmelen, Frank, Vladimir Lifschitz, and Bruce Porter, editors. Handbook of Knowledge Representation. Elsevier Science Ltd, 2008.
   This comprehensive handbook covers foundational methods, specialized representations, and applications in knowledge representation, authored by leading experts. It is organized into three parts addressing logic, reasoning methods, and practical AI applications

.

Brachman, Ronald J., Hector J. Levesque, and Raymond Reiter, editors. Readings in Knowledge Representation. Morgan Kaufmann, 1985. A classic collection of seminal papers foundational to the field of knowledge representation.

Russell, Stuart J., and Peter Norvig. Artificial Intelligence: A Modern Approach. 4th ed., Pearson, 2020. Although broader in scope, this widely used AI textbook contains substantial sections on knowledge representation and reasoning.

Brachman, Ronald J., and Hector J. Levesque. Knowledge Representation and Reasoning. Morgan Kaufmann, 2004. A landmark textbook providing a clear and detailed introduction to symbolic knowledge representation and reasoning techniques

   Reiter, Raymond. Knowledge in Action: Logical Foundations for Specifying and Implementing Dynamical Systems. MIT Press, 2001.
   Focuses on logical foundations for representing dynamic knowledge and reasoning about actions.

   Minsky, Marvin. The Society of Mind. Simon & Schuster, 1986.
   A seminal work presenting a theory of knowledge representation based on interacting agents.

   Sowa, John F. Knowledge Representation: Logical, Philosophical, and Computational Foundations. Brooks/Cole Publishing, 2000.
   Provides a broad overview of knowledge representation from multiple perspectives including logic and conceptual graphs.

Theses and Dissertations

   Brockmans, Sara. "Metamodel-based Knowledge Representation." 2007. Karlsruhe Institute of Technology. http://digbib.ubka.uni-karlsruhe.de/volltexte/1000007322

Schlobach, Klaus Stefan. "Knowledge Discovery in Hybrid Knowledge Representation Systems." King’s College London, University of London, 2002

Khor, Sebastian Wankun. "A Fuzzy Knowledge Map Framework for Knowledge Representation." Murdoch University, 2007

ONLINE COURSES

University of Freiburg

https://gki.informatik.uni-freiburg.de/teaching/ss18/krr/lecture.html

   Knowledge Representation for Intelligent Applications — University of California, Berkeley
   Covers theory and practice of building knowledge graphs, language models, semantic web technologies, and reasoning with structured information. Emphasizes hands-on Python labs and projects.
   URL: https://www.ischool.berkeley.edu/courses/info/290/kria

Knowledge Representation — Vrije Universiteit Amsterdam Focuses on formalisms like Description Logic, Default Logics, Argumentation Frameworks, and Probabilistic Graphical Models with practical reasoning tool implementation. URL: https://studiegids.vu.nl/en/courses/2024-2025/XM_0059

Knowledge Engineering with Semantic Web Technologies (MOOC) — openHPI (Hasso Plattner Institute) Teaches fundamentals of Semantic Web, ontologies, linked data, and knowledge representation on the Web. Includes practical examples and additional in-depth content. URL: https://open.hpi.de/courses/semanticweb2015

Open Source Tools for Knowledge Representation — Central European University (CEU) A training session highlighting open-source knowledge representation languages and data structuring for research optimization. Part of CIVICA Research Open Social Science Training. URL: https://library.ceu.edu/open-source-tools-for-knowledge-representation

TOOLS https://webprotege.stanford.edu WEB PROTEGE https://logictools.org/index.html SKOS to OWL https://www.heppnetz.de/projects/skos2owl/