Public Safety Knowledge Graph using Bilstm- Attention-CRF and Multi-Source Data

Abstract

Public safety has always been a hot topic in the world, yet related data are stored in various forms of structures and have a huge amount. The poor characteristics of these huge amounts of heterogeneous data bring great difficulties to its application. For the purpose of applying these data conveniently and efficiently, knowledge graph was adopted. Public safety knowledge graph can help organize public safety data by an organized and properly utilized form, so that it can be used more optimally in public safety related retrieval, question and answer systems, etc. In this paper, unstructured, as well as structured public safety data were used for constructing, with the support of BiLSTM-Attention-CRF model. To be able to reveal this public safety knowledge graph in an understandable way, Neo4j database was used to store public safety knowledge graph.