Research And Design Of The Subjective Question Auto-Assessment System

With the rapid development of computer technology and communicationtechnology,the computer has been applied to all areas of people’s lives. In the field of education, the computer can not only assisted teaching,but also be assisted evaluation. The system is research of the Intelligent Marking System for subjective questions.The system is based on Natural Language Processing(NLP).It contains five main sections:sentence segment,word segment,syntax analysis,keyword extraction and sentence similarity computing.It adopts ICTPROP and Forwoard MM to get sentence words.Then,It uses LTP analyze the grammatical structure of the sentence.And It extracts keywords by filtering the grammatically-partial words in sentence.By analyze the synatic tree identifying sentence components.After that,It extracts keywords of each part.At last,It calculates the similarity of the sentence.The key points of the system is word segment and sentence similarity computing.The focus of the system is the word segment and the word similarity computing. The word similarity computing is based on hownet.The System increase specialized vocabulary of network protocol,and it can identify specialized vocabulary by FFM. Word similarity computation is based on HowNet word similarity algorithm. HowNet word similarity algorithm includes two parts:calculation of concept similarity computing and obligations of the original similarity.Through the analysis of the meaning of the original tree, The system improves the existing algorithms,so it more suitable for the system.