Abstract

Representing world knowledge in a machine pro-cessable  format  is  important  as  entities  and  their descriptions  have fueled  tremendous  growth  in knowledge-rich information processing platforms,services,  and  systems.   Prominent  applications  of knowledge  graphs  include  search  engines  (e.g.,Google Search and Microsoft Bing), email clients(e.g.,  Gmail),  and  intelligent  personal  assistants(e.g.,  Google  Now,  Amazon  Echo,  and  Apple’s Siri). In this paper, we present an approach that can summarize  facts  about  a  collection  of  entities  by analyzing their relatedness in preference to summa-rizing each entity in isolation. Specifically, we gen-erate informative entity summaries by selecting: (i)inter-entity facts that are similar and (ii) intra-entity facts that are important and diverse.  We employ a constrained knapsack problem solving approach to efficiently compute entity summaries.  We perform both  qualitative  and  quantitative  experiments  and demonstrate that our approach yields promising re-sults  compared  to  two  other  stand-alone  state-of-the-art entity summarization approaches.