Research On Building A Knowledge Graph Of Food Safety Cases And Events Based On Enhanced BERT

Knowledge graphs,as a novel means of knowledge representation and reasoning,provide new ideas for research in the field of food safety.Knowledge graphs represent and store information such as entities,attributes and relationships in food safety cases and events in an efficient way,improving the efficiency of food safety knowledge expression and communication.At the same time,through intelligent reasoning and analysis based on knowledge graphs,potential risk factors in cases can be quickly identified,possible food safety events can be predicted,and corresponding measures can be taken in a timely manner to improve food safety assurance.Therefore,the establishment of food safety knowledge graphs has important theoretical and application values.This paper addresses the construction of a knowledge graph of case events in the field of food safety.The construction of the knowledge graph is carried out through two main parts,firstly,named entity recognition,which extracts entities with specific meaning in the text,and then establishes the relationships between entities based on the text.The specific research in this paper is as follows:(1)To address the problem of the large number of types of entities in the food safety domain and their strong specialisation,a named entity recognition model incorporating food safety specialised vocabulary is proposed,which can effectively recognise specialised vocabulary in the food safety domain,and combines a two-way long and short-term memory network model to learn the contextual features of food safety texts and a conditional random field model to constrain the recognition results to improve the recognition accuracy of the model.A food safety named entity recognition dataset is constructed based on the model features,and the model is trained and validated.The experimental results show that the model effectively improves the recognition accuracy of named entities in the food safety domain.(2)Based on the characteristics of food safety text,an entity relationship extraction model that can be integrated into existing food safety expertise mapping data is proposed.A food safety entity relationship extraction dataset is constructed,and the effectiveness of the model is verified by comparing it with other models on this dataset.(3)Based on the above model,the adjudication documents are constructed into a food safety case knowledge map and the food sampling records data are constructed into a food safety event knowledge map,which together are combined into a food safety case event knowledge map,on which the food safety expert seminar system is designed.This system enables the detection of expert statements,the automatic extraction of keywords related to food safety from the expert statements,the search for relevant entities and relationships in the food safety case and event knowledge map,and the real-time display.By building a food safety knowledge map,the foundation is laid for effective food safety research and reasoning.