Fusion of simplified entity networks from unstructured text

To explore the fusion of unstructured text, the concept of entity networks has been proposed in many systems to facilitate analysis of entity relationship. Using appropriate visualization, entity networks could illustrate the relations of extracted entities. In this paper, we shall present an approach for combining multiple entity networks. The entity networks are derived from documents and hence, it is also a scheme for document-to-document fusion. Using an entity and relation extractor, an entity network could be built. However, such a system typically generates many entities and links from a single document, and some of them are trivial. With complicated entity networks, it is difficult and less effective to perform fusion. As such, a simplification algorithm is introduced to construct a simplified version for each entity network. Incrementally, by using our fusion algorithm, the entity network could be updated by adding a new network. A combined entity network is constructed, in which one could get the key concepts and their relationship. It is a summary of the multiple documents. The examples in this paper demonstrate the purport and effectiveness of the simplification and fusion scheme.