Addressing Information Overload through Text Mining across News and Social Media Streams

The state-of-the-art in topic detection and tracking, structured summarization and news recommendation has moved to alternative document representations beyond keywords, in an attempt to utilize the available metadata in the form of timestamps, entities, topics, categories, author information, sentiment, political stance. Yet, despite the availability and the advantages of social metadata, only a few methods have attempted to utilize social annotations for document representation beyond social posts. This report briefly introduces the use of social annotations for news in near-real-time settings and answers the question - Are the social annotations useful for tackling single-domain multiple-document tasks in the news domain, such as topic detection and tracking or story structure extraction?