Research On Mongolian And Chinese Automatic Question And Answer In The Field Of Five Livestock Based On Knowledge Graph

With the continuous development of digital society and the continuous growth of artificial intelligence technology,the application scenarios in various fields have become more and more complex.In the field of natural language processing,the research on automatic question-answering in specific fields has also become in-depth and diverse.Ordinary search engines return a series of articles based on keywords,while the Question and Answer system gives more accurate answers through the analysis of natural language.It is a more advanced and accurate search engine,which reflects the intelligence of the computer.The knowledge graph aggregates a large amount of structured semantic information of knowledge,can model all kinds of things,and describe all kinds of concepts and the relationship between them in detail.It is the cornerstone on the road to cognitive intelligence and promotes the development of artificial intelligence.Q ＆ A based on the knowledge graph can make use of the relationship between entities,provide high-quality knowledge sources,understand the semantics of the input questions,complete operations such as query,reasoning,and concept retrieval,help improve the efficiency of retrieval information,clarify the semantic information behind users’ queries,and return the correct answer.However,in the field of Mongolian information processing.It not only lacks search engines similar to Baidu Encyclopedia,but also lacks the application of the knowledge graph technology.In order to solve the above two problems,it is necessary to study the question answering system in the field of Mongolian.By realizing the more advanced way of retrieving information,it can deeply understand the semantic knowledge of Mongolian,promote cultural exchanges among all ethnic groups and promote the construction of the common spiritual home of the Chinese nation.（?）（Five Livestock）in Mongolian refers to horses,cattle,camels,sheep,and goats.In Mongolian,the special words related to the five livestock are very rich and colorful and have distinctive national characteristics.For example,the category,gender,age,hair color and temperament of the five livestock have detailed names.There are many common names or dialect local words.The five-livestock words form a complete semantic field.This thesis aims to realize Mongolian and Chinese automatic question and answer in the field of five livestock based on the knowledge graph（Mongolian and Chinese Five Livestock Question and Answer,MCFLQA）.The specific research contents are as follows:（1）Acquire and process five-livestock data sets.Translate Mongolian data to form Mongolian and Chinese bilingual structured data,which is convenient and efficient to extract entities,relationships,and attributes.（2）Construct the knowledge graph of Mongolian and Chinese in the field of five livestock.Extract entities,relationships,and attributes in a structured way,construct the knowledge graph,and use graph database to realize visualization.（3）Identify the five-livestock entities in the question.AC multi pattern matching algorithm and text similarity algorithm are combined to form a five livestock entity extraction method to support accelerated entity recognition,question type matching and other operations.（4）The multi-classification model determines the query intention.Collect Mongolian and Chinese bilingual interrogative words to improve the accuracy of sentence meaning analysis,and manually collect 1013 questions for the input of the classification model.TF-IDF and Embedding features are used to build classification models,they include Linear SVM,Nonlinear SVM,Nave Bayes Model,Logistic Regression,Random Forest,XGBoost,LightGBM.The experiments show that the Linear SVM classification model established by embedding features has the best classification effect,it can judge the user’s intention by classifying the question text.（5）Complete Mongolian and Chinese bilingual automatic question answering based on the knowledge graph.Realize the question answering system through question sentence understanding,semantic matching,answer retrieval,and other operations.Randomly choose 300 Mongolian and Chinese bilingual Q ＆ A pair of five categories to test the system performance,the average accuracy was 89.2%.It provides a strong platform for the dissemination of Mongolian five livestock knowledge and disease questions and answers.