COMP30412: Knowledge Representation (2008-2009)
The Web is one of the largest and most diverge bodies of knowledge ever. Strikingly, it is also both a relatively coherent information artifact, a distributed computation system, and a medium for human interaction on a wide scale. While much of the content of the Web originates in databases and other well modeled forms of data, it is primarily exposed in the form of HTML. In recent years, Web Services, REST services, XML, and other Web 2.0 technologies have striven to augment the "human oriented" web with a "program oriented" Web, e.g., a Web of Data.
Among these competing approaches, the Semantic Web is distinctive in reimagining the Web not as a Web of Data, but as a Web of Knowledge. The key idea of the Web, of course, is distributed hypertext. The key idea of the Web of Data is a distributed datastore. The key idea idea of the Semantic Web is a distributed knowledge representation
The field of Knowledge Representation (KR) lies at the intersection of (at least) Artificial Intelligence and Information Management. KR is the attempt to provide rich representations of the world and various things in it that supports building programs that are sensitive to the world via these representations. KR has been used to build expert and diagnostic systems, speech recognizers, games, automated planners, etc.
The Semantic Web is where KR and the Web collide. This course will explore the aftermath.
In particular, we will focus on various logic based formalisms for knowledge representations including their design and use. We will look at attempts to represent various parts of commonsense and scientific knowledge, as well as the use of KR for conceptual modeling in information systems. We will explore how to bring such formalism (and associated representations) into a Web context, with special attention paid to the unique challenges thereby raised.
We will also analyze the problems and promises of KR through discussion of some of the seminal articles of the field.
The course will provide students with an understanding of logic and logic-based knowledge representation formalisms, their theoretical and practical aspects, some relevant reasoning services, and how these are used to support modelling. It will also discuss various issues in knowledge acquisition and engineering with an emphasis on realistic applications.
A student completing this course unit should:
1. Have an understanding of knowledge representation, its motivations, applicability, advantages and pitfalls (A)
2. Have knowledge and understanding of some basic KR formalisms namely first-order and description logics. (A)
3. Have knowledge and understanding of the way in which automated reasoning can be used to help with modelling. (A and B)
4. Have mastered the basic range of techniques for building knowledge representations using standard tooling (B, C and D)
Assessment of Learning outcomesLearning outcomes will be assessed by examination. Outcomes 1 and 4 will also be assessed via exercises and laboratories during the course.
Contribution to Programme Learning OutcomesA1, A5, B1, B3, C4
Topic covered include:
First order logic: syntax, semantics, proof theory, and applications
Description logics: syntax, semantics, proof theory, and applications
Ontologies and ontology engineering
Commonsense and scientific representation
Selected papers and technical reports will be distributed during the lectures.