Implementasi Metode Terms Frequency-Inverse Document Frequency (TF-IDF) dan Maximum Marginal Relevance untuk Monitoring Diskusi Online

Abstract

The application of social media during the process of teaching and learning especially in online discussion forum is gradually increased. Neverthelles, the spreading of out of scope discussion that trigger the emergence of negative debates breaks the communication etic code in online discussion. This push forward the increasing of admins or instructurs rules in monitoring and controlling the discussion activity during the forum session.. By applying TF-IDF and Maximum Marginal Relevancy methods a software apllication is developed to monitor the discussion online activity. The list of Text Processing Phase including The sentences breakdown, case folding, tokenizing, filtering and stemming are conducted to extract the document posting from the instructurs as well as members comments. Then, TF-IDF, Query Relevance and Similarity values are calculated. By applying Maximum Marginal Relavancy, the optimal extraction of documen summary is provided to reduce the sentences redudancy and rangking output. The comment which value is zero (0) that based on the comparison of summary between document posting and members comments will be classfied as “Unfeasible” and recommended to be eliminated. As the result of accuracy, blackbox and UAT testing in one of lecture topics this application is success in monitoring the activity of online discussion with compression value 50% and level accuracy is 76,67%.   Hence the discussion forum as one of tool in incerasing the teaching and learning quality can be optimaized accordingly.