A text mining model for hypothesis generation

This paper presents a tool to detect links between two topics across documents (e.g. two individuals). We interpret such a query as finding the most meaningful evidence trail across documents that connect these two topics. We propose to use link analysis techniques over the extracted features provided by Information Extraction Engine for finding new knowledge. A concept-association- graph based approach was proposed which combines text mining, information retrieval and link analysis techniques. Experimental results on the counterterrorism corpus demonstrate the effectiveness of our algorithm. Specifically, the algorithm generates ranked concept chains where the key terms representing significant relationships between topics are ranked high ¹.