Multi-document Summarization Based on Cohesion with Disambiguation

Abstract

The mass information of Internet boosts the requirement for quick and accurate methods of information acquisition. To fulfill the requirement for high quality multi-document system, this paper investigates using lexical cohesion as a model of multiple-documents written in Chinese to generate an indicative, moderately fluent summary. The method constructs lexical chains with polysemant disambiguation and identifies strong chains first. An unique disambiguation method that combined the sense definitions in the dictionary with the expanded context of the words is presented. Then significant sentences that extracted from each document are merged, and redundant information are recognized and removed. Finally, the summary is generated in chronological order. Evaluation results show that the performance of the presented system is obviously better than that of the baseline system.