Abstract:

In order to help exploratory search users discover important and valuable concepts for domains, and understand aspects of the concepts and comprehensively understand the problem domains, based on a concept usage model, concept importance and how problem aspects affect concept importance are analyzed to attain a method supporting multiaspected concept discovery. Such a method considers mutual affects between concept importance and problem aspects of concepts, leading to a simultaneous analysis on the two factors. Experimental results show that the proposed method could reduce search interactions by 48.8% and could improve concept occurrences by 24.3%.

Key words: exploratory search, concept discovery, rank cluster, concept importance