Paper Reading #19: Tell me more, not just "more of the same"

Comments:

TBD

Reference Information:

Title: Tell me more, not just "more of the same"

Author:  Francisco Iacobelli, Northwestern University, Evanston, IL, USA; Larry Birnbaum, Northwestern University, Evanston, IL, USA; Kristian J. Hammond, Northwestern University, Evanston, IL, USA

Presentation Venue: IUI '10 Proceedings of the 15th international conference on Intelligent user interfaces

Summary:

The researchers of this paper are attempting to creating a new system of search that will automatically provide the user information that is related but not identical to the initial search results. They call this system Tell Me More. With the search information that is provided, it will take that seed story and search for similar related stories. However, instead of spouting back the same results that match exactly the search criteria, it will take additional text excerpts based on secondary search criteria that provide information above and beyond the primary data the user requested.

The Tell Me More system consists of five core modules to accomplish its task. These modules are Content Gathering, Content Filtering, Text Analytics, Difference Metrics, and Presentation. The Content Gathering module gathers documents from search engines and other user defined sources. The Content Filtering takes the material returned from the first module and filters out things that are basically identical to the seed material. The Text Analytics module analyzes the material for unique information related to the seed topic. The Difference Metrics determines from the analyzed material what is the unique information to include. The Presentation module categorizes and ranks the resulting analyzed material.

Discussion:

This system sounds like it could provide a unique take on the now classic Google search. I always find a lot of repeat in the results I get and if I don't find what i want in the first 10 results, I tend to modify my search parameters in order to provide myself with a different result. With this system, it sounds like all of that additional effort I put into my searching would be handled for me automatically.

I could really do with an improved search. What we have right now is pretty good, but I can see that there is room for improvement. The system that is being proposed in this paper seems like one way in which the standard search mechanism can be improved and make the process richer, if not more efficient. The speed is currently fine, depending on the connection you have. But making the results more meaningful would be a good improvement for our daily search needs.