Internet learning behavior power-law distribution probability modeling method in large-scale online education

Abstract

The invention discloses an Internet learning behavior power-law distribution probability modeling method in large-scale online education. The method includes the steps of 1, obtaining a primary influence factor set Fori of Internet learning behaviors B in line with power-law distribution; 2, adding factors the correlation coefficients of which are greater than a set correlation coefficient threshold valve r into a candidate influence factor set Fcon; 3, adding factors with causal relations with the Internet learning behaviors B into a final influence factor set F; 4, disclosing a probability model assumption of Internet learning behavior power-law distribution; 5, creating a probability model representing the probability that learners take part in a learning activity on a certain day; 6, simulating the learning processes of all learners within a course period. According to the method, a universal Internet learning behavior probability model and a learning process generation algorithm are created, the method is suitable for various Internet learning behaviors and capable of achieving prediction on the learning behaviors of the learners in a follow-up period after courses on the basis of the model, correspondingly a teaching plan is dynamically adjusted, and remedial measures are taken in time.

Description

technical field [0001] The invention relates to the field of network learning behavior analysis, in particular to a power-law distribution probability modeling method for network learning behavior in large-scale network education. Background technique [0002] With the maturity of network communication technology and the increasing scale of distance education, the analysis of network learning behavior has gradually attracted the attention of researchers in related fields. At present, the relevant research on the analysis of online learning behavior in the field of online education (e-Learning) is mainly divided into five categories: research on influencing factors, research on regulatory mechanisms, research on interactive behavior, research on learning styles, and research on learning behavior modeling. [0003] In 2013, Romero C, López M I, and Luna J M et al published Predicting students' final performance from participation inon-line discussion forums in Computers&Educat...

AI Technical Summary

Problems solved by technology

However, the research in this literature pays attention to correlation analysis, but ignores causality analysis, and fails to correctly understand the difference between correlation and causality, that is, correlation is neither a sufficient condition nor a necessary condition for causality, which leads to the accuracy of factor analysis. lower, lower reference significance

At the same time, most of the existing research is based on small-scale e-Learning, so the excavated learning rules or established models are not applicable to large-scale e-Learning

Images