Paper Reading #20: Supporting Exploratory Information Seeking by Epistemology-based Social Search

Comments
Jorge Perez 
Stephen Morrow


Reference Information
Title: Supporting Exploratory Information Seeking by Epistemology-based Social Search
Authors: Yuqing Mao, Haifeng Shen, Chengzheng Sun
Where/When: IUI 2010


Summary
People often have problems creating proper keywords and evaluating search results for current search engines, which are part of exploratory information seeking (EIS). This paper suggested a new way of allowing users to search with hopes of making it faster and easier. This system is based on epistemology social searching, which reuses and redefines other users' contributions, allowing users to share web searches.

A prototype, Baijia, was created, in which users submit a query and selected search pages for it, which are then added to the epistemology. Search pages can be ranked and commented by other users dynamically in real-time. Related queries can then be presented to the user, allowing information to be presented quickly. Users can subscribe to a epistemology, and if it is refined by others, they will receive notifications about it.

An example of Baijia, showing epistemology generation and refining.

An experiment was conducted using AOL query logs as a base, then evaluating searches using Baijia. Overall, this system showed better ranking performance over AOL. This was the only feedback provided, but it was stated that user studies are in process.


Discussion
I was a little confused about what an epistemology was, but this sounds like it could greatly improve web searches, especially for those that do not know how to form proper query words or that do not know what kinds of result pages might be good. And let's face it, there are many people that still ask questions of a search engine instead of just using keywords. I also think it was cool that they gave a social aspect to searching, by allowing users to rank and refine others' searches. I would like to see what further user studies show about this system to see if it would actually be good or not.