Paper: HCAMiner: Mining Concept Associations for Knowledge Discovery through Concept Chain Queries

ACL ID C10-3008
Title HCAMiner: Mining Concept Associations for Knowledge Discovery through Concept Chain Queries
Venue International Conference on Computational Linguistics
Session System Demonstration
Year 2010
Authors

This paper presents HCAMiner, a system focusing on detecting how concepts are linked across multiple documents. A tra- ditional search involving, for example, two person names will attempt to find documents mentioning both these indi- viduals. This research focuses on a dif- ferent interpretation of such a query: what is the best concept chain across multiple documents that connects these individuals? A new robust framework is presented, based on (i) generating con- cept association graphs, a hybrid content representation, (ii) performing concept chain queries (CCQ) to discover candi- date chains, and (iii) subsequently rank- ing chains according to the significance of relationships suggested. These func- tionalities are implemented using an in- teractive visualization paradigm...