Analysis and Architecture Design of a Large-Scale Event-Centric Knowledge Graph System for Dispute Resolution

Exploiting natural language processing and machine learning techniques, knowledge graph (KG) can transform disorganized raw data into structured knowledge. Recently, KG has been applied in various fields, such as intelligent search, question answering systems and so on. Most of them are usually entity KGs, which can be denoted by triples including head entities, tail entities and relations between them. This kind of KGs generally focus on named entities, e.g. people, organizations, places. With the development of the Internet and social media, large amounts of news spread rapidly and widely. Both academia and industry have particularly concerned of how to accurately mine the information and acquire the details of related events from lots of news. Besides that, events of disputes or conflicts may bring threats to social security. As a result, deeply analyzing and mining information of conflicts and disputes can effectively improve the ability of social governance. In this context, we introduce an event-centric knowledge graph to assist solving the problem. In this paper, we first discuss the architecture of our proposed event-centric knowledge graph for dispute resolution, which is suitable for large-scale data processing. Then we introduce the methods and modules to construct the proposed architecture. For each module, we discuss the details of our building methods. To this end, we capture a complete view of the proposed system.