Research On Automatic Knowledge Graph Construction Techniques For Fire Emergency Management

In recent years,the ability to respond to emergency management of fire incidents has become increasingly important.Knowledge graphs can be used to establish physical,temporal and spatial relationships between entities in the emergency response domain,providing supplementary support for government emergency command and dispatch and post-disaster data analysis.However,as the scale of data in the field of fire emergency management is very large,the manual construction of knowledge graphs is not only slow but also inefficient.To address these problems,this paper proposes the automatic construction of knowledge graphs in the field of fire emergency management.It is of great significance to maintain public safety by providing a perfect and scientific data system for fire emergency management workers,thereby improving the efficiency of fire emergency management.The main research content of this paper is as follows:（1）In order to extract important entity knowledge from textual information in the field of fire emergency management,this paper proposes a deep learning based CNN＿Bi LSTM＿CRF named entity recognition model for automatic entity extraction.The local and global information of the text is combined by CNN and Bi LSTM,and then the named entity recognition is completed by sequence annotation using CRF.The experimental results show that the combined CNN＿Bi LSTM＿CRF model proposed in this paper has a better entity extraction effect compared with the base model.（2）This paper proposes a remotely supervised PCNN＿Bi LSTM＿Attention relationship extraction model to achieve automatic relationship extraction.By combining segmented pooled CNN and Bi LSTM,the local and global features of the text are automatically learned,and then the Attention mechanism is introduced to reduce the problem of incorrect annotation and noise caused by the remotely supervised learning process.Experimental results show that the combined PCNN＿Bi LSTM＿Attention model proposed in this paper is more effective in relation extraction compared with the basic model.（3）As the domain knowledge lacks a certain architecture,the knowledge needs to be standardised and constrained through ontology files.This paper therefore proposes a relational database to ontology based conversion algorithm,which automatically generates OWL ontology files from the knowledge of the underlying database through the conversion algorithm.The generation of ontology files makes the knowledge in this domain more standardised and systematic.