| "Answering Online" is one of the important parts of network teaching system. While teaching in network, teachers can't answer all the questions asked by students in time, owing to limitation of time and vigour of teachers, deficiency in face-to-face communication between teacher and students, huge amount of student compare to limited amount of teachers, which gives rise to difficulty and deficiency in mastering knowledge for students. Thus we should establish a set of high-effective and natural language based answering system, which not only can make students' questions to be answered quickly ,but also can lighten the work pressure of the teachers and can put main vigour of the teachers on the keypoint to develop introspection instruction. In this thesis, author analyse and make a conclusion that weak chain of the research of answering online system based naturallanguage——Chinese Information Handle Technologe, comparing to researchsituation of answering online system both at home and abroad. It is very difficult that realize answering online system based natural language, such that the questions asked by users can't be dealed with properly because analyse of questions only was limited to simple depatching keywords and can't reach deep semantic feature; that the identical and different meaning of question is difficult to distinguish because the way and keypoint of question wasn't considered at all;that computing the similarity of questions is deficient, question wasn't ordered by the similarity,and can't present the most relavant answer.Therefore, this system drew lessons from the experience and deficiency,and design the questions following by the users, seek and expand the question of database resources. Experts need add the professional keywords related the question to question database while expand the question database. The keywords was devided into first keywords, second keywords, etc. according to professional keywords' function in the sentence. While users asked, the answering online system match the professional keywords in resource database to the question asked by user firstly,that is to check if the question asked by user contain the first professional keywords or contain the second professional keywords, or so on. If contained, the number "1" was returned,or the number "0" was returned. The number was multiplied by the weight of respective keywords again,then added all the number.At last the total number is the computing similarity of the question asked by user,the system will order the related question in database resource by the total number,and present them to each user.The user will select and confirm the question they want to ask,then the answer of user's question will be return by the answering system.The user will assess the answer given by the answering system,the expert/teacher will modify the database of the questions and answers.In addition,this answering online system combine both the person-computer interaction model and the person-person interaction model as well,which not only supported the quick answer by answering online system based natural language,but also increased the interaction between the teachers/experts and students,such as message box used for interacting the information between the teachers and students.And the system increase the expanding function of the unknown question,that is that the system will add the question to the unknown question database automatically while there wasn't the question asked by user in the question database resource.The system increase the user-assessment function,where users give the assess-score according to the answer presented by the system,and the total score would be feedbacked to experts/teachers.The teachers modified the answer according to the score feedbacked by the users.And the system still add the FAQ function,which not only reflect the FAQ question,provide rethink instruction for teachers,but also provide example for users,and make them know which question is keypoint. |
PDF Full Text Request