Improvement Of Data Mining Algorithm And Application For Teacher-Student Interaction Platform

At present, the interaction between teachers and students in the network environment is not limited by time and space, which has brought great convenience.But the interaction is inefficient due to lack of close organization and effective supervision. The thesis discusses how to improve algorithms in data mining and applies the data mining technology for the teacher-student interaction platform to improve the communication efficiency between teachers and students through interacting network platform.The thesis analyses the strengths and weaknesses of classic Apriori algorithm for association rules mining.We employ0-1matrix to describe the original set of database transactions, put forward a inclusion algorithm for0-1matrix to seek the maximal frequent item sets and improve the Apriori algorithm through reducing the frequency of pruning and database scanning.The thesis focuses on the typical advantages and disadvantages of clustering algorithm based on k-means clustering algorithm, introduces and analyses semi-supervised k-means algorithm, and find that exception marked data can cause the algorithm to converge slowly. To solve this problem, we put forward semi-supervision of k-means algorithm to deal with the exception marked data, look for more suitable initial cluster centers, speed up the process of convergence and improve the performance of the semi-supervised k-means algorithm.By applying improved algorithms to teacher-student interaction platform, we can mine the students’browsing data and identify the associated pages to guide students to browse and improve learning effectively. The improved semi-supervised k-means algorithm is used to sum up key points.By clustering the messages’data, we can find out contents that students are interested in and guide teachers to update the web site.