Wednesday, April 13, 2011

Paper Reading #21: Epistemology-based Social Search

Comments:
Comment 2

Reference Information
Title: Supporting Exploratory Information Seeking by Epistemology-based Social Search
Authors: Y. Mao, H. Shen, C. Sun
Presentation: (Conference Paper) IUI'10, Febrary 7-10, 2010, Hong Kong, China

Summary: In this paper the authors present a system called Baijia, an epistemology-based social search solution for the problem of finding the proper keywords and evaluating results in search engines. The system reuses and refines previous successful searches to provide users with successful, accurate and desired information.

The system uses packages of information derived from a number of resources such as search processes, queries, results, ranking, annotation, comments, etc, in order to provide the user with a successful search process even if he entered keywords that are not as relevant or are vague. Basically, the system has an epistemology repository, and every time a user adds a query, it is added, and can then be linked to previous epistemology records, comments, web pages, etc. all with the purpose of presenting the user information about previous searches.

Even though they have not yet performed user studies, their experimental evaluations have showed that such a system outperforms conventional search systems.

Discussion: Even though I think this may be a good idea to improve search results, I don't think the paper did a great job in explaining the system. I suppose such an application can be useful for people who are trying to find out about something, but they don't really have the words to describe it. Also, if users have the capability of commenting on the results of the search, that can allow users to warn each other about certain websites that are not useful.


3 comments:

  1. This does seem like a good idea, so far I have read multiple papers about this type of searching. Some of the other papers also try to include recommendations based on your search history and on what other users have found relevant to a similar search.

    ReplyDelete
  2. The social aspect of the design is very interesting. I'm curious how well this would develop if widely used.

    ReplyDelete
  3. I would really like to see this kind of system in use to see how well it actually works. I would love for it to work well, but it's very reliant on other people contributing to the system. Wikipedia seems to function just fine by people maintaining it, so a search system like this one might be able to work the same way.

    ReplyDelete