Event Knowledge Graph Extracted from Intelligence Text

For the problem of knowledge extraction and application of battlefield intelligence text, this paper proposes to use Hierarchical Event Knowledge Graph to build a model of battlefield events knowledge, and presents a framework consists of Specific Temporal Graph and Abstract logical Graph. Based on new methods of Event Recognition and Extraction, we obtained information of events and relationships from battle intelligence text using Wargaming data as the original corpus. The mapping between facts and concepts is also achieved by the method of Trigger Recognition. Then we propose a method of assignment of edge weights for the Event logical Graph, and construct the hierarchical structure. Finally, the visualization of knowledge was achieved using Visjs. It is shown that the method can effectively extract event knowledge from specific texts, and support downstream knowledge reasoning tasks for Assisted Decision-making.