Semester Offering: InterSem
 

The use of the Internet to market products has provided companies with the unprecedented ability to personalize information, products, and services to individual consumers. Personalization has been shown to be an effective tool in boosting sales and increasing customer loyalty.

 

Theory and techniques for personalizing information and products. Discussion of techniques for building user models and using them in tailoring or searching for information or products, including collaborative filtering, content-based filtering, rule-based filtering, and multi-attribute utility theory. Applications from the standpoint of the consumer and of the producer. Examples include creation of personal newspapers, recommending music or movies, and helping users to choose among large sets of complex products, e.g. computers.

 

AT06.20 Knowledge Management and Information Retrieval or AT02.10 Data Structures and Algorithms or Consent of Instructor.

 

I.          Representation and Elicitation of User Preferences
1.    Structuring Objectives
2.    Preference Structures and Value Functions
3.    Preferential Independence
 
II.        Machine Learning Techniques
1.    Decision Tree Induction
2.    Backpropagation Neural Networks
3.    Theory Refinement
 
III.       Personalization Techniques
1.    Collaborative Filtering
2.    Content-Based Filtering
3.    Rule-Based Filtering
4.    Computer-Assisted Self Explication
5.    Information Visualization
 
IV.      Applications
1.    Targeted Advertising
2.    Information Filtering
3.    Product Search and Selection
 
V.       Survey of Current Research
 
VI.      Case Studies of Commercial Sites

 

Course packet consisting of relevant sections from textbooks, papers from current literature, and lecture notes.

 

T. Mitchell:
Machine Learning, McGraw-Hill, 1997.
 
R. Keeney and H. Raiffa:
Decisions with Multiple Objectives, Cambridge University Press, 1993.
 
H. Kautz:
Recommender Systems: Papers from the 1998 Workshop, American Association for Artificial Intelligence, 1998.
 
W. Hanson:
Principles of Internet Marketing, Chapter 7, South Western College Publishing, 1999.
 

 

Communications of the ACM, Special Issue on Personalization, vol 43, no 3, Aug 2000.
Journal of the ACM
Artificial Intelligence
Machine Learning
Decision Support Systems
AI Magazine

 

The course is organized in two parts: (i) lecture and (ii) reading, presentation, and discussion of papers from the current literature. Papers are selected from recent conferences. Students are required to read all papers and to write two 1-page paper analyses each week. Each student selects one paper to present in front of the class. The intention is to give students practice with the essential skills of critical reading, analysis, written expression, and oral presentation.
 
The final grade will be computed from the following constituent parts:
 
Mid-semester (15%),
Final (25%),
Homework and paper analyses (15%),
Paper presentation (10%), project (25%), and
Project presentation (10%).
 
Closed-book examination is used for both the mid-semesterand final exam.