Text Mining Analysis in the Log Discussion Forum for Online Learning Recommendation Systems

These studies aim to determine the degree of similarity between student and lecturer posts. The level of similarity then we validated using a lecture note. This research is a continuation of the research that we are currently doing related to the development of smart LMS with one of the supporting features of the recommendation system. Using the analysis in the forum log post discussion of this research was carried out through several stages. The first stage, the random selection of 5 classes taken by the forum data, the second stage, carried out text mining analysis from the posts of students and lecturers, the third stage, analyzing text validation using lecture notes that have been used as data sets in the form of corpus. This study uses the doc2v algorithm with vectorization. The results of this study found that the percentage of similarities between lecturers' posts, students and lecture notes only reached 49% of the target we expected at least 80%. Because discussion forums are a substitute for face-to-face sessions on face-to-face learning. On the other hands this study found that the similarity of discussion between lecturers and students on discussion forums had a significant influence on student learning outcomes (assignment) and this reinforces the need for a system of recommendations for online learning.