ACQUISITION OF ANECDOTES FROM THE WEB FOR FREE CONVERSATION SYSTEMS

A free conversation system that can freely talk with users is a typical human-like computer that can enrich our daily life. As knowledge used for the free conversation system, this paper proposes a method to automatically acquire anecdotes of various people from Web. First, passages that are related to a given person are retrieved by search of relevant web pages and segmentation of the obtained pages. Then, each passage is judged whether it is the anecdote by several rules based on linguistic features of the anecdotes. The experimental results showed that the precision of our method was increased by 25% comparing to the baseline.