Thursday, June 30, 2011

Your best might not be good enough -- re-thought on search engine

From our recent submission (Prantik is the first author):

The primary objective of a web-search engine like Google is to find and index the best content entry (or entries) for some given keywords, where best is usually based on some form of popularity through link analysis algorithms [1]–[3]. If the target of the search is either well-known/established (e.g., Newtons Laws of Motion) or clearly controversial (e.g., Californias Proposition 8 on Marriage Protection Act), then Google (or Wikipedia) can provide very nice information entries covering the target topics. However, if a topic is either dominated by a small school of thoughts or still on the converging process from various random ideas, the model of ranking based on popularity might not effectively help an individual in quickly identifying what he is really looking for. For a good example of ourselves, for the keywords of “Future Internet Design”, we had to flip five search results pages under Google before we can find the first link (http://goo.gl/KrO4B [4], results retrieved on June 20, 2011) about our Davis Social Links project [5].

Information is also about who have written, read, and shared about the information, and more importantly, how those social entities (e.g., readers/authors) are related to ‘us’ in interpreting and ranking that particular piece of information. For ‘Future Internet Design’, those who are related to our
research team socially should have a better chance to find out about our particular thought on this converging topic. We believe that the integration and correlation of sharable information content on the Internet and the on-line social informatics (i.e., relationships and interactions) is a possible direction to help our society in resolving the conflict between the quantity and the personalized quality during an information search process. The process of a user sharing URL(s) with friends leads to an integration of the web graph with the social network graph. The integration of the two networks helps in the growth of a trusted web where random pieces of articles from the web find support in the social network and user attempts at finding relevant articles for queries can boost from URL(s) that have been already read and shared by their friends for their relevancy.

No comments:

Post a Comment