Conversation Timeline: Tracing Tête-a-Tête in News Media

Abstract

Today there is vast amount of information/news available on the World Wide Web in the 'form of news articles, blogs and microblogs. Everyday the news media is filled with news L about various entities commenting about different events. Thus each event generates a series of conversations between different entities of interest. It would be very interesting for a user to discover how a specific entity has conversed with other entities in the news media about various events happened within a certain time period.We propose a novel visualization technique, "Conversation Timeline" that chronologically displays the conversations of a given entity with other unknown entities which have been reported in news media. We have used news headlines to extract these conversations since news headlines contain the key idea of the comments made by an entity. We use relation extraction algorithm to retrieve conversations from the text. We propose a pre-processing step that syntactically alter the headlines so as the relation extraction module can generate better quality relations. We believe that a Conversation Timeline will give a better understanding about how an entity interacts with other entities of interest. lv