Research on answer generation model of coal mine safety knowledge question answering system
With the gradual improvement of national and coal mining industry′s requirements for emergency management of coal mines,higher requirements have been put forward for learning coal mine safety knowledge.Therefore,an intelligent question answering model for coal mine safety knowledge is established.The effective study of coal mine safety knowledge are crucial to ensure the personal safety of coal mining enterprise staff and prevent the occurrence of coal mine safety accidents.The answer pair data can be generated automatically based on RoBERTa-wwm algorithm,and the question types are defined and the question answering pairs are labeled by obtaining and analyzing the original text data of coal mine safety knowledge.By combining with RoBERTa-wwm and UniLM,the point mutual information and adjacent entropy are used to discover new word expansion domain dictionaries,propose an automatic question answering pair generation algorithm,and construct the question answering pair dataset of coal mine safety training knowledge,so as to solve the problem of question answering dataset in coal mine safety knowledge system.By introducing the question similarity mechanism,an answer generation strategy is proposed for unanswerable questions and irrelevant questions,and the answer generation model based on the question similarity mechanism is constructed to focus only on answerable questions and improve the reasoning ability of the model.The experimental results show that the proposed answer generation model of the coal mine safety knowledge question answering system can effectively identify the unanswerable and irrelevant questions,and can provide knowledge support for the coal mine enterprise staff,so as to improve the safety training and learning effect of the coal mine enterprise staff to the greatest extent.
万方数据
维普
