Complementary information retrieval for cross-media news content

We propose a novel way of integrating cross-media news content such as TV programs and Web pages to provide users with complementary information. Our method can be used to search for cross-media news content that complements news items in which a user is particularly interested, i.e. complementary content that provides more detailed information or a different perspective on the topic rather than just similar information. First, we propose a novel content-representation model called a “topic-structure” model. A topic structure consists of a pair of subject and content terms. Subject terms are the dominant terms in a news item and content terms are terms that have strong co-occurrence relationships with the subject terms. Using this topic structure, we search for information related to the news item in which the user is interested from the perspectives of content, context, and media complementation. We also describe an application system that enables a TV news program to be presented concurrently with complementary news Web pages, providing the viewer with an easy way of acquiring more details about a news topic from different perspectives.