Design And Implementation Of Intelligent Question Answering System Based On Service Matching

With the development of artificial intelligence,especially the deepening of information technology research,how to extract the knowledge that user needs from the massive information has become a popular research direction,followed by the birth of a large number of information retrieval system,traditional way of using the browser search engine has been difficult to meet the needs of users,the emergence of intelligent Q & A system makes it easier for users to get useful information.But now the research on the intelligent Q & A system is still in the early stage,can only answer some common questions,the process of deal with question are also relatively simple,the answer to the professional areas such as finance,medical care,electricity and other issues is difficult to get the results of customer satisfaction.This paper summarizes the existing technologies of intelligent Q & A system,there are three difficulties in system implementation,the first is how to build the corresponding knowledge base and search method according to the problem domain,the second is how to accurately resolve the query semantics from the user's questing,the third is how to match the query semantics to the corresponding search rules,and then get the answer from the knowledge base.The research of this paper is mainly for these three points;the specific contents are as follows:The construction of the knowledge base is based on the domain knowledge of the query problem,for the domain knowledge has partial structure,strong correlation characteristics,in this paper,the domain knowledge is stored in the relational database,and the concept,attribute and relation of the data are extracted by the ontology,in this case,the content of the ontology is only the domain meaning and related relation of the entity in the knowledge base,and does not include the whole knowledge base instance,which provides great convenience for the establishment of the ontology and the maintenance of the knowledge base.Because the domain knowledge-related query has the characteristics of strong semantic understanding,the corresponding entity or attribute description of the problem is clear.This paper abstracts the query structure of the query problem into the corresponding query service,to establish the retrieval method of the knowledge base.Q & A system is generally through the voice to ask questions,the existing Q & A system is default that voice input method identification is correct,but in fact for some professional problems,because of the existence of special nouns in Chinese and English,has many new words and lack of corresponding vocabulary knowledge base,coupled with the existence of spoken language is not standard and other reasons,resulting in the input method of voice recognition error situation.In this paper,we propose a method to correct the text after voice recognition based on the vowel,and use the conditional random field to annotate the entity.Through this method,we can accurately identify the semantics of the question.According to the level of domain knowledge and relevance,this paper presents an improved service matching method,this method can judge the matching degree of the service parameter concept through the domain conceptual relationship,and use the non-necessary attributes of the service description to improve the accuracy of the service query and reduce the scope of the services to be matched.Consider the case where the query has a contextual association,this article save the results of the query in a structured form,and according to the corresponding query conditions to determine whether the use of context to the next query.Based on the above research,this paper proposes a framework of intelligent Q & A system based on service matching.The framework divided into three modules: knowledge combing,problem analysis,service retrieval and query context management module.In order to verify the practicality of this framework,this paper applies the framework in the "National Economic Data Q & A System",and the whole application process has achieved good results,which proves that the framework proposed in this paper is effective and feasible.