Building and exploring an enterprise knowledge graph for investment analysis

Abstract

Full-fledged enterprise information can be a great weapon in investment analysis. However, enterprise information is scattered in different databases and websites. The information from a single source is incomplete and also suffers from noise. It is not an easy task to integrate and utilize information from diverse sources in real business scenarios. In this paper, we present an approach to build knowledge graphs (KGs) by exploiting semantic technologies to reconcile the data from diverse sources incrementally. We build a national-wide enterprise KG which incorporates information about 40,000,000 enterprises in China. We also provide querying about enterprises and data visualization capabilities as well as novel investment analysis scenarios, including finding an enterprise’s real controllers, innovative enterprise analysis, enterprise path discovery and so on. The KG and its applications are currently used by two securities companies in their investment banking and investment consulting businesses.