Correlation mining for web news information retrieval

Abstract

In this chapter, we focus on the problem of correlation mining in news retrieval. To this end, we present a framework of multimodal multi-correlation news retrieval, which integrates news event correlation, news entity correlation, and event-entity correlation simultaneously by exploring both text and image information. The proposed framework enables a more vivid and informative news browsing by providing two views of result presentation, namely, a query-oriented multi-correlation map and a ranking list of news items with necessary descriptions including news image, title, central entities and relevant events. First, we preprocess news articles using common natural language techniques, and initialize the three correlations by statistical analysis about events and entities in news articles and face images. Second, considering the sparsity of the known event-entity correlation, an algorithm of Multi-correlation Probabilistic Matrix Factorization (MPMF) is proposed to reconstruct it with joint consideration of the three correlations. Third, the result ranking and visualization are conducted to present search results. Experimental results on a news dataset collected from multiple news websites demonstrate the attractive performance of the proposed solution.