Entity summarization with high readability and low redundancy

Abstract

The development of the World Wide Web has triggered substantial growth of knowledge graphs (KG). Research into using KGs for intelligent applications has increased significantly. A KG describes facts about entities using RDF triples, and an entity description may contain a large number of triples. In applications where entity information is presented directly, entity summarization is required to prevent user information overload and to fit the presentation capacity. Here, the task is to select the most representative subset of triples from the rich entity description. In this paper, we propose an innovative entity summarization method, which we refer to as ESSTER, to generate summaries with both high readability and low redundancy. The proposed method combines structural and textual features. The importance of a triple is measured based on its structural features in the KG. The text readability of a triple is measured based on n-grams in a text corpus, and redundancy in a set of triples is measured by logical reasoning, numeric comparison, and text similarity. Entity summations is modeled and by combining these three measures and solved as a combinatorial optimization problem. We conducted experiments and compared the proposed method to six existing methods on two publicly available datasets of manually labeled summaries. Experimental results demonstrate that the proposed method achieves state of the art results.