"You gotta make your own kind of music
Sing your own special song"

(Mama Cass Elliot)

My research goals are to explore and develop computational models and techniques to help users better access and retrieve more relevant information while doing their jobs more efficiently and effectively. To achieve my research goals, I have focused on inter-disciplinary research in three main areas, specifically: user modeling, information retrieval and human factors. The fundamental idea behind my research is to capture information dynamically about a user and use it to improve retrieval performance and user performance. In my approach, I combine the system-centered approaches and user-centered approaches by taking a well-established evaluation framework in information retrieval while exploiting the strengths of formal knowledge representations in artificial intelligence for building the model.In addition to the main research stream, I am also interested in computer supported cooperative work (CSCW), human computer interactions (HCI), human factors (HF) and distributed computing (DC). Below is the list of my papers classified into several main research topics and the summary of my dissertation.

The final version of my dissertation can be downloaded here with its appendix can be downloaded here. And the slides for my defense can be downloaded here.

Papers classified by research topics

Cognitive user model supported intelligence analysts : ICIA05a , ICIA 05b. HCISS05

Text Summarization : HCISS06

Geospatial and Distributed Information Retrieval :SPIE05, SPIE06

User Models for Information Retrieval: ICAI00, IAT01, UM03, AAAI Doctoral Consortium 2004, AAAI MOO Workshop 2006

Empirical Evaluations of Adaptive User Model: UM03, AH04

Human Facrors: HFES03, HFES04

Collaborative Filtering: UAI99, AAAI Workshop 99, AAAI Workshop 98

Misc: AIPS98, IC2000,

Summary of Dissertation

I study the problem of employing a user model for information retrieval in which knowledge about a user is captured and used for improving retrieval performance and user performance. In my dissertation, I improve retrieval performance and user performance for information retrieval by capturing user intent dynamically through analyzing information from retrieved relevant documents and through combining it with the elements of an information retrieval application. I use semantic representation and decision theoretic principles for building this model. I develop the first unified evaluation framework to assess the effectiveness of the model using the selected collections from the information retrieval community and quantitative usability testing method from the human computer interactions community. The results have shown that our approach retrieves more quality documents in the initial run and performs competitively with the best traditional approach for relevance feedback from the information retrieval community (Ide dec-hi) after getting feedback. In addition, we have done a pilot study with human intelligence analysts. The results have shown that the user modeling approach tracked the analyst’s intent and adapted to each analyst’s search styles better than an off-the-shelf commercial keyword-based retrieval application.