Research And Application Of Open-Domain Automatic Question Answering System Based On Deep Learning

Open domain automatic question answering system aims to provide information acquisition based on natural language,it has a wide range of applications in search engines,intelligent personal assistants and so on.At present,the popular automatic question answering system mainly includes community question answering system and knowledge base question answering system.The community question answering system finds historical questions similar to user questions from the content community and returns their answers to the user as the answer to the user's question.The knowledge base question and answer is mainly to transform the natural language into the structured query statement,and then query the knowledge base such as the knowledge graph to obtain the answer to the current question,and the answer is usually an entity.In community question answering system,because user questions and existing questions are relatively short,it is difficult for the retrieval model based on keyword matching to achieve good matching accuracy.Because the vocabulary and the query rule set are written manually,the knowledge base question answering system becomes difficult to maintain and expand with the expansion of the database.Therefore,there are some automatic question answering systems based on deep learning have emerged to alleviate these problems.In this paper,the reading comprehension technology of natural language is studied,and an automatic question answering system is built.The main contents are as follows:In this paper,two machine reading comprehension models based on deep learning are proposed,one of which is an extractive reading comprehension model.the model finally extracts one word or several consecutive words from a given text as the final answer to the question.The other model is the generative reading comprehension model,which combines the given text and finally generates the answers from the default dictionary.Both models encode natural language text directly and generate answers.They are end-to-end models.In this paper,an automatic question answering system based on reading comprehension is implemented by combining the above model with the existing search engines.The experimental results show that the comprehensive performance of the system is better than that of the existing automatic question answering system,and it has good generalization performance,it has good results in the open field.