Li Jia, Zhang Pengzhu, Li Xinmiao, Jihie Kim
In order to stimulate student thinking and reduce service response time, the authors propose an online recommendation method that mines relevant information from discussion logs. The proposed method consists of six steps, namely technique term hierarchical tree building, task term extraction, document segmentation, feature extraction, topic identification, and document score calculation. Two experiments are designed to evaluate the proposed system. In the first experiment, three different document recommendation algorithms (i.e., TF-IDF, TF-IDF + topic filtering, TF-IDF + LSA + topic filtering) are compared, and the TF-IDF + topic filtering algorithm gets the best performance. In the second experiment, the proposed system is applied to one semester of online discussion, and the proposed system leads to better discussion participation as well as a higher level of perceived usefulness and ease of use. The results indicate that moderated relevant discussion record can be mined automatically, and providing such information can stimulate student's thinking and discussion.
