Sentiment analysis in teaching evaluations using sentiment phrase pattern matching (SPPM) based on association mining

This research proposes a new sentiment analysis method called sentiment phrase pattern matching (SPPM). The analysis model extracts the responses and comments from discussions that are posted in a teaching evaluation system in the form of open-ended questions and allows student respondents to provide feedback to their teachers on factors that affect teaching and studying in a classroom. The proposed method consists of three main phases: (1) collect feedback data and perform tokenization via the Teaching Senti-Lexicon; (2) analyze sentiment analysis phrases by SPPM, which is based on the association mining method and integrated with sentiment phrase frequency by using forward bigram traversal, for separating the many phrases from teaching feedback sentences; and (3) sentiment analysis based on sentiment scores from the Teaching Senti-Lexicon. The objective of this research is to obtain feedback from open-ended questions automatically via the proposed method for sentiment classification and to determine the best classification of the responses to the open-ended questions within educational attitude contexts by classifying attitude contexts as positive or negative. Moreover, SPPM is compared to others classifier algorithms. The results indicate that the SPPM method achieves the highest accuracy of 87.94% compared to the other classifier algorithms. In addition, SPPM achieves precision, recall and F-measure values of up to 92.06, 93 and 92.52%, respectively. The main contribution of the proposed model is that it determines the most effective strategy for improving teaching based on students’ opinions.