Intelligent Technology Of Subjective Question Marking Based On The Similarity Of Text

At present,in all kinds of tests at home and abroad,most of the common subjective questions,such as short answer and composition,are corrected by traditional manual methods.The quality of the scores has strong human factors,and the efficiency is low,so it is difficult to achieve objective and fair.With the rapid development of artificial intelligence and natural language processing technology,more and more researchers pay attention to the intelligent marking technology of subjective questions,which has become an important research topic in the field of educational technology.In recent years,the development of artificial intelligence technology,especially the wide application of neural network in text similarity calculation,not only improves the accuracy of Chinese text similarity,but also provides the technical possibility for the development of subjective intelligent marking and improving the accuracy of marking.Among them,the proposition of dependent syntactic structure solves the differences between grammatical and syntactic structures of Chinese texts.It not only improves the accuracy of similarity calculation of Chinese texts,but also promotes the development of intelligent marking technology for subjective questions.Therefore,based on the above research,this paper designs an improved text similarity calculation method and applies it to subjective intelligent marking to improve the accuracy of subjective marking.The main contents of this paper are as follows:(1)Aiming at the application of dependency relation and synonym word forest in text similarity,a semantic similarity calculation method based on the combination of dependency relation and synonym word forest is designed.The method extracts the relationship paths of two texts by dependency relation,and calculates the semantic similarity of the relationship paths between two texts based on the synonym forest.The semantic similarity between the two texts is calculated by combining the relational path and the synonym forest.The experimental results show that the algorithm improves the accuracy of text similarity calculation.(2)A text similarity calculation method based on long-term and short-term memory network is designed,and Early Stopping mechanism is added to the experimental training process to avoid over-fitting and non-convergence.Experiments show that the similarity calculation method also performs well in subjective intelligent marking.(3)Designed an intelligent scoring model of subjective questions based on the above algorithm,and used a monthly examination paper of a high school history subject to test a large number of subjective questions.The experiment shows that the algorithm can control the difference between the intelligent scoring of subjective questions and the score given by teachers in a relatively reasonable range.In summary,the intelligent marking technology of subjective questions based on text similarity can reduce the workload of teachers,improve teaching efficiency,and improve the accuracy of intelligent marking of subjective questions.