Research On Q&A System Of Chinese Ancient Books Based On Knowledge Graph

The vast number of Chinese classics is the memory carrier of the Chinese nation and the crystallization of ancient sages’ wisdom.And The Translation of Among Flowers(Hua Jian Ji)is one of the shining pearls of the Chinese classics.However,the content of Hua Jian Ji is concise and the authors like to use allusions and current events to express their emotions,which makes it difficult to read and understand today.Therefore,if we can learn about the knowledge beforehand,and then savor the "words、sentences、links and pieces",it will be easier for us to understand the connotations of Hua Jian Ji and provide more nourishment for our spiritual world.With the rapid development of information technology,users choose search engines to search answers when acquiring knowledge.However,using search engines to search will suffer from problems,such as retrieved information that is not the right answer or complicated and redundant,which makes it difficult for users to obtain knowledge accurately.On the one hand,Q&A system can semantically parse the question to quickly understand user’s intention,on the other hand,it simplifies the result and provides user with high-quality answers directly.And Q&A System based on Knowledge Graph realizes the systematization and integration of fragmented knowledge by storing knowledge in the form of triple,which provides a highquality data source for Q&A system and further facilitates user’s queries.Therefore,this paper takes Hua Jian Ji as an example,focuses on information extraction and knowledge graph construction,namely,builds named entity recognition model and knowledge graph,and also builds a Q&A system based on knowledge graph.The system can parse the question input by user,query knowledge graph and return answer,helping user to obtain knowledge faster and more accurately,which has certain practicality.The main work in this paper is as follows:(1)A named entity recognition model is constructed.Since tunes of poems are different,position of entities will change accordingly.And poem’s corpus that have been labeled is relatively small,so the corpus needs to be manually labeled according to the writing background of poem and the emotion that author wants to express.Therefore,this paper artificially annotates Hua Jian Ji as the dataset for named entity recognition and proposes the AM-NER named entity recognition model to recognize entities.Finally,comparison experiments and ablation experiments are designed to verify the superiority of the model.(2)A Hua Jian Ji knowledge graph is constructed.Since there are different structures of knowledge on the Web and a large amount of knowledge exists in unstructured data.So,different technological approaches are needed for knowledge acquisition and storage.Therefore,this paper,firstly,acquires the knowledge of Hua Jian Ji through crawling technology based on the characteristics of the crawling website.Secondly,the manually labeled entities and the knowledge from different data sources are subjected to the processes of knowledge fusion and knowledge processing,and the knowledge is stored in the form of triple to realize the construction of Hua Jian Ji knowledge graph.(3)A Hua Jian Ji Q&A system is constructed.Based on the above research results,this paper realizes a Hua Jian Ji Q&A system based on knowledge graph,and the specific process includes: question analysis,question processing and visualization display.