A Classification Framework For Disambiguating Web People Search Result Using Feedback

This paper is concerned with the problem of disambiguating Web people search result. Finding the information about people is one of the most common activities on the Web. However, the result of searching person names suffers a lot from the problem of ambiguity. In this paper, we propose a classification framework to solve this problem using an additional feedback page. Compared with the traditional solution which clusters the search result, our framework has lower computational complexity and better effect. we also developed two new features under the framework, which utilized the information beyond tokens. Experiments show that the performance can be improved greatly using the two features. Different classification methods are also compared for their effectiveness for the task.