ChronoSeeker: search engine for future and past events

In this paper, we propose an on-demand search engine called ChronoSeeker, which allows users to find past/future events based on their interest. Our goal is providing a search engine which can collect as many future/past events as possible relevant to user's query in obtaining various future scenarios considering both predictions and histories. Two technical issues are treated, (1) efficient search method for event information and (2) accurate filtering method for removing noises from search results. To search for event information effectively, our system expands a user query by some typical expressions related to event information such as year expressions, temporal modifiers and context terms. To remove noisy information, we selected five types of features for a machine learning technique to classify candidates into event information or not. Our experiment showed that filtering performance achieved an 85% F-measure, and that query expansion can collect dozens of times more CEs than those without expansion.