Concept Expansion Based Answer Summarization

This paper presents a concept expansion based unsupervised method for answer summarization in the integer linear programming framework. The expanded question is linked with its associated a saver sentences by the concepts they have in common. An objective function can be built using their mutual concepts. A readable summary is obtained by maximizing the objective function. The method proposed is tested on the DUC 2007 summarization data and Yahoo! Answer data. Experiments show the method performs better than traditional methods.