Common Sense Search

Ambiguous web search queries often only return results that correspond to the most popular meaning. This is a product of the way that search engines rank page results. This makes it difficult to find information about less popular meanings. This project attempts to provide a solution to this problem by allowing users to search through semantic clusters formed from an original search query. These clusters show a semantic mapping of the search space that is unrelated to the relative popularity of the various meanings. It is also hoped that this type of system would be useful for general search as well as mapping out ambiguous search spaces. Introduction and Motivation A common problem encountered by web users performing searches is dealing with mediocre search results resulting from ambiguous search terms. When a user searches using ambiguous terms, it is difficult for search engines to produce accurate results for all users. For example, when a user enters the search term “triangle” it is impossible to know whether the user is referring to a generic three-sided figure, a musical percussion instrument, or perhaps the Bermuda Triangle. If such a search is done, typically the most popular subject overwhelms the search results even though the other subjects are semantically just as related to the query. Most of the time this sort of behavior is acceptable, but it makes it more difficult for users to search for more obscure information. This is related to the effect that search engines have on the popularity of pages: popular pages become more popular (since they are easier to find with search engines and, therefore, more people will link to them), obscure pages continue to be obscure (no one can find them to link to, so their page ranks remain low). It would be useful to have a mechanism by which search results could be clustered into separate subject groups and then the user could decide which subject he or she wanted to pursue. This clustering mechanism should not be based on subject frequency, but instead should rely on more meaningful semantic grouping. Copyright c © 2008, American Association for Artificial Intelligence (www.aaai.org). All rights reserved. It is the goal of this project to leverage the power of OpenMind/CommonSense to provide useful semantic clustering. This repository of human knowledge may be able to produce usable clusters which will help disambiguate user queries.