Combined analysis of news and Twitter messages

While it is widely recognized that streams of social media messages contain valuable information, such as important trends in the users’ interest in consumer products and markets, uncovering such trends is problematic, due to the extreme volumes of messages in such media. In the case Twitter messages, following the interest in relation to all known products all the time is technically infeasible. IE narrows topics to search. In this paper, we present
experiments on using deeper NLP-based processing of product-related events mentioned in news streams to restrict the volume of tweets that need to be considered, to make the problem more tractable. Our goal is to analyze whether such a combined approach can help reveal correlations and how they may be captured.