The Study Of Text Mining And Application In Network Teaching Platform

The task of text mining is discovering content features of texts. It involves many subjects, for example, Natural Language Processing, Machine Learning, Data Mining and so on, and has been widely used in various fields. There is so much unstructured/semi-structured data in the Network Teaching Platform, which can be effectively managed and organized by utilizing text mining.This paper gives an introduction of research at home and abroad concerning text mining and subjective questions’ automatic assessment firstly. And subsequently, on the basis of Single Similar Degree, the author measured the matching degree of students’answers and keywords of standard answers and then achieved the automatic assessment functions for Short Answer Question.But compared to Short Answer Question, the Essay Question is more inclined to be affected by the complexity of Chinese semantics, and therefore, the according algorithm has been designed-Similarity Assessment based on semantics. In this paper, the author calculated the semantic similarity between each word with Hownet in the first place, and then sentence, paragraph and text successively by the model of Similarity Matrix and the Maximum Sequence. If the K-nearest neighbor is applied to Automatic Assessment, it will probably exert harmful influence on performance on account of diversity of sample size and density. And it is the reason why the Fixed Weighing Factor is raised to weaken the negative impact Finally, the author performed a series of experiments to test the algorithms and then implemented the Automatic Assessment for Short Answer Question and Essay Question in the Task Model of Siyuan Network Teaching Platform.