Research And Implementation Of Chinese Automatic Response Question Answering System Based On Sentence Similarity Measure

Massive Open Online courses(MOOC)is a new learning form based on the network and intellectual mobile technologies.As a totally new approach to study,it not only promoted the modern distance education,but also brought the traditional education,especially higher education a huge revolution.Because of geographical separation,though students study on MOOCs,it's difficult for them to communicate with teachers directly.The answering of questions,as an indispensable link in the process of learning activities,its effect will directly affect the learning quality.At present,most MOOC learning platforms answer questions through forums and social platforms like MSN,learners are unable to get the teacher's guidance immediately when they encounter the problems,as a result it is difficult to achieve good learning results.At the same time,the most important characteristics of the MOOC course is its large-scale and openness,it's impossible for teachers to answer all the questions from students.So how to find the answer rapidly and accurately is a challenge to improve the user experience of MOOC learning platform.The function of question answering track system is similar to search engine,it provides users with related answers to a question and supports learners ask questions by natural language.There is no need to split the questions into keywords,the system returns a concise and accurate answer instead of some relevant web pages.It can be more accurate than the search engine to find out the answer,meet the demand of the user's retrieval.Applying question-answering technology to MOOC learning platform and answering problems of learners automatically within a certain range can not only help learners to solve problems and follow up study,improve the timeliness and reusability of the resources,but also help teachers to analysis and find the learner's weaknesses and improve the deficiency of the teaching.To this end,this paper designed QA-system based on the frequently asked questions.This system realized automatic asking and answering through the calculation of similarity of questions.In this paper,it designed a measurement method based on semantic dependency relationship,and combined with characteristics of sentence length,words form,meaning and other factors to measure the similarity between sentences.The experimental results show that the method can well reflect the relationship between the sentences,it is a feasible and effective method and improve the accuracy of the system.This system has high response accuracy,strong use value and broad application prospects.In this paper,the main research work is listed as follows:(1)Analyzing the concept,theory and method of the related technologies in question answering system and sentence similarity measures by researching the existing literature.(2)It proposed a sentence similarity calculation method according to semantic dependency parsing,it breaks the bondage of the syntactic sentence structure and accesses to the deep semantic information,it also evaluate sentence similarity according to the sentence semantic relevance of various language units.(3)It realized the sentence similarity calculation method based on the same words,sentence length,semantic dictionary,vectors,edit distance,semantic dependency parsing,proposed a new method to measure sentence similarity and combine multiple information,experiment has been done to find out the parameter in algorithm which makes the system correct.(4)It designed and developed an automatic QA-system,completed the automatic response by the sentence similarity calculation method designed in this paper,auxiliary teaching,realized application value.