The Research And Application Of WEB-log Mining In MOODLE

In this Paper, we have done some research about data mining algorithms, and used R language and SQL query to analyze Moodle logs. And we have achieved a number of theories or thoughts for Application design. We mainly focused on 4 points: Analysis of Learning Path, Analysis of Learning Process, Prediction of Leaning Effect, and Recommendation for Course Resources. For detailed, we mainly did the following work:1. First, we summarized the Learning System on construction, running state, and data for years, which was built in 2012 and based on Moodle. We collected and summarized a lot of requirements about Content Design and The data analysis during the years. And we spent a couple of months on learning R with common algorithms, and SQL query.2. And then, we exported a large number of user logs, and summarized the log data. And we have done two data analysis tasks using SQL query. One was Analysis of Learning Path, and another was Analysis of Learning Process. According to the analysis results, we proposed the concept about learning path design: Structuring the Contents and learning path linearing. And then, we designed the data structure on Moodle. Next, we designed the method of learning process analysis, quantization, and visualization on that structure.3. And lastly, we used R and the methods of regression, CART and k-mean to do two experiments. One was Prediction of Learning Effect, which predict student's score using logs. And another was Recommendation for Course Resources, which used k-mean to cluster the logs by time, and designed the steps to generate Recommendation data.