 |
|
|
Title: Evaluation of Attention Profiles
Introduction: This paper presents a research proposal for an evaluation study of attention profiling, a form of user modeling that captures a person’s attention on the Internet. Our proposal explores two options for the study: 1) profiling users’ web browsing behaviors; 2) profiling online music listeners’ taste. The first option is a qualitative case study of a small group of users of Cluztr, a Firefox browser plugin. The second option is a quantitative, cross-sectional study of a larger group of users of Pandora, an online music recommender. Both options involve an evaluation of relevancy and accuracy of the recommendations provided by each respective service. Our objective is to investigate whether attention profiling accurately matches people’s interests and information needs. We will also examine factors that impact the accuracy of attention profiles. The outcome of this research could lead to the adoption of attention profiling in services that provide information retrieval (IR) and information filtering (IF).
Downloads:
 Report (336KB)
Presentation (388KB)
 Poster (488KB)
|
 |
|
|
|
 |
| |
Special Topics Presentation
|
|
|
|  |  |
|
 |
|
|
View Notes
Title: Google PageRank Algorithm
Introduction: Many of us are fascinated with the success of the Google search engine since its inception in 1998. What was the landscape of the World Wide Web back then? What made the Google search engine unique among all of the other search engines at the time? The answer lies in Google’s PageRank algorithm. In 1998, Sergey Brin and Lawrence Page, two Stanford University doctorate students, presented their search engine prototype called Google at the Seventh International World Wide Web conference (WWW98) in Brisbane, Australia. The Google search engine is based primarily on the PageRank algorithm, which is different from the three traditional information retrieval (IR) models: boolean model, vector space model, and probabilistic model. In their own words, Brin and Page defined PageRank as "an objective measure of its citation importance that corresponds well with people’s subjective idea of importance" (Brin & Page, 1998). This paper presents an overview of the unique problems in IR on the web, design goals of the Google search engine, description of the PageRank algorithm, and system architecture of the initial prototype.
Downloads:
Report (165KB)
Presentation (806KB)
|
|
|
|
|
|
|
|
|
|
|
|
| Week |
Topic |
Date |
| Week 1 |
Overview of the course; Introduction to IR and IF systems |
August 28 |
| Week 2 |
Introduction to IR and IF systems (cont.); Research paradigms in IR |
September 4 |
| Week 3 |
IR models: Set theoretic models - the Boolean model |
September 11 |
| Week 4 |
IR models: Algebraic models - the vector model |
September 18 |
| Week 5 |
Retrieval evaluation; System vs. User-Centered evaluation: test collections (e.g. TREC); recall, precision, relevance, and alternatives measures |
September 25 |
| Week 6 |
Query languages: keyword-based querying, pattern matching, query protocols (Z39.50, WAIS) |
October 2 |
| Week 7 |
Query operations: user Relevance feedback; query expansion and reformulation for the vector model (similarity thesaurus; statistical thesaurus) |
October 9 |
| Week 8 |
IR on the web: Hypertext; search engines; metadata; markup languages |
October 16 |
| Week 9 |
Information representation; Text operations: lexical analysis, stemming, compression, thesauri |
October 23 |
| Week 10 |
Indexing (data structures); natural language vs. controlled vocabulary; alternative retrieval techniques (e.g. natural language processing, citation processing) |
October 30 |
| Week 11 |
User modeling in IR |
November 6 |
| Week 12 |
User modeling in IF; Taxonomies of IF |
November 13 |
| Week 13 |
User modeling in IF. Dimensions of user models. Characteristics of the user and of the user need |
November 20 |
| Week 14 |
IF performance; Profile quality evaluation; Personalization: ethical issues; Project Presentations / Evaluation |
December 4 |
|
|
|
|
|
|