Research On Key Technologies Of Intelligent Question Answering System Based On Semantic Understanding

The rapid development of big data and artificial intelligence has led to a sharp increase in the scale of network data.Traditional search engine returns unordered list of matched web pages,and the results searched contain a large amount of irrelevant information,which is no longer sufficient for the current users.The intelligent question answering system integrates natural language processing,knowledge management and other technologies to understand query intent,and returns the exact answer to the user in natural language.But most of the existing question answering systems use the common corpus as a knowledge source,the data is redundant and the dimension is single.At the same time,influenced by synonyms and polysemous words,the query intent cannot be reflected by the input questions,which causes the term mismatches and some other errors.The rapid development of semantic Web provides a good solution to the above problems,more and more researchers try to apply the key technologies such as ontology construction and semantic query in the semantic Web to the question-answering system,in order to solve the problem of lack of semantic understanding.This paper combines semantic Web with question answering system,using ontology as the knowledge source,describing information in the semantic level and knowledge level.The semantic understanding and knowledge reasoning are added to solve the problem of poor question matching degree and fuzzy user query intention in traditional question-answering system,aiming at optimizing the semantic comprehension ability and to provide accurate and comprehensive question and answering services.Main innovation theory and research results have been proposed as following:(1)Sentence semantic similarity calculation of multi-feature fusion.In the calculation of sentence similarity,the structure information and semantic information of sentences are considered synthetically.The length features,morphological features,and word order features are extracted,and the analytic hierarchy process are used to calculate the structure similarity.The weights of the concepts in the ontology are given to calculate the semantic distance,and then the semantic similarity is described based on the semantic distance between concepts.And finally,the weighted fusion method is used to calculate the overall similarity.Compared with the traditional cosine similarity algorithm and the similarity algorithm based on editing distance,the similarity calculation method of multi-feature fusion obviously improves the calculation accuracy.(2)Semantic-based question query expansion.The query keyword sequence is obtained after the processing of user query,and the query keyword is mapped to the domain ontology according to the concept of occupancy rate.The improved minimum spanning tree algorithm is used to generate the minimum query spanning tree,and the query spanning tree is extended by the effective path.The semantic similarity is used to filter and rearrange the extended words set,and finally the keyword extended words set of the user query is generated.The query intent is described more clearly from the semantic level,and the ontology-based query efficiency is improved.Experimental results show that the semantic based query extension obtains a higher F-measure than the no-extension method and the keyword extension method.(3)The building of an intelligent question answering system based on semantic understanding.Based on the background of book,the book domain ontology,the database of frequently asked questions and the database of greetings questions is constructed.The question answering strategy of the combination of the frequently asked questions and semantic is designed to realize the accurate answer of the user's questions.The system also provides a visual interface to display the similarity calculations,query expansion and data sources,verifies the feasibility and effectiveness of the above algorithm.