NIA: System for News Impact Analytics

The analysis of news impact on people is relevant to a variety of applications, ranging from monitoring product and companies reputations, to stock market prediction. Therefore, it is important to understand the underlying mechanisms which affect the propagation of news and drive the evolution of sentiments in one way or another. In this demonstration paper, we describe NIA, a system that identifies and describes news events that caused changes of sentiments. NIA is based on a novel framework for a complex news event modeling, which is capable of detecting time and importance characteristics of events by only observing a time series of news articles publications, and then correlating this data with a time series of sentiment shifts. The operation of our system is summarized as follows. First, we apply a deconvolution to recover the time, longitude, importance and impact of news events. Second, we compute a sentiment time series, e.g., by monitoring sentiments for positive or negative bursts, and coherently analyze sentiment and news time series, automatically determining their time lag. Third, we evaluate the corresponding news articles for a time interval of interest and extract the essence of what happened. Finally, we present the selected news time series to the user, as well as several more correlated stories, which could have affected sentiments as well, proposing to interactively explore their connections.