ICOMS 4725 Knowledge Representation and Reasoning

  Description: The field of Knowledge Representation is primarily concerned with encoding general world knowledge in a formal system of computation. A central part of Artificial Intelligence, Knowledge Representation has links to many fields including logic, computer science, linguistics, cognitive psychology, and cognitive science. 

This course will focus on central KR methodologies, such as first-order logic, extensions to first order logic (modal and default logics), inheritance, semantic networks, frame systems, and production systems. 
We will study these methodologies from two vantage points: 
that of the designer of commercial AI systems, and that of the researcher aiming to develop a system capable of commonsense reasoning. 
We will focus on applications in the medical, legal, and insurance domains.    
 
 

Course Outline:

 
Note that there will be no classes on March 31 or on April 7. One class will be made up during midterm week; one class will be made up during reading week.
Class
No.
Topics/Chapters Covered
I. January 20
 General knowledge representation issues; 
motivating applications (medical, legal, business); overview of first-order logic.  
 
II. January 27
 First-order logic, continued; developing ontologies; limitations of first-order logic (lack of expressivity, lack of inferential power); knowledge engineering issues.
READINGS TO PREPARE: Davis: Chapter 1, Chapter 2: pp. 27-52, Morgenstern: drafts of Chapters 1 and 2 of Advanced Reasoning  
 
III. February 3
 Reasoning about time; temporal logics; introduction to modal logics.
READINGS TO PREPARE: Davis: Chapter 2: pp. 52-66, Chapter 5; Morgenstern: draft of Chapter 3 
 
IV. February 10
 Modal logics of knowledge and belief;
possible-world semantics;
deontic logics for legal reasoning.
READINGS TO PREPARE: Davis: Chapter 2: 59-75, Morgenstern: draft of Chapter 4, other readings to be announced.  
 
V. February 17
 Syntactic logics of knowledge and belief;
quotations, paradoxes, and resolutions;
planning; knowledge and planning. 
 
VI. February 24
 Semantic networks; description logics;
standard inheritance; inheritance tools and applications. 
 
VII. March 3
 Default and nonmonotonic logics; closed-world assumption;
circumscription; belief revision. 
 
VIII. March 10
 Nonmonotonic inheritance networks; algorithms. 
 
IX. March 24
 Formula-augmented networks; applications to medical reasoning and the insurance industry. 
 
X. April 14
 Frame languages; Bayesian networks. 
 
XI. April 21
 Production systems, expert systems. 
 
XII. April 28
 Multiple-level knowledge representations (e.g., combining semantic and Bayesian networks);
formal and informal mappings between different knowledge representations.
 
XIII. May 5
 Open issues in knowledge representation.