Opinion Target Extraction for Student Course Feedback

Opinion Target Extraction for Student Course Feedback Janaka Chathuranga, Shanika Ediriweera, Pranidhith Munasinghe, Ravindu Hasantha and Surangika Ranathunga Department of Computer Science and Engineering University of Moratuwa, Katubedda 10400, Sri Lanka {janaka.13, shanika.13, pranidhith.13, ravindu.13, surangika}@cse.mrt.ac.lk Abstract Student feedback is an essential part of the instructor - student relationship. Traditionally student feedback is manually summarized by instructors, which is time consuming. Automatic student feedback summarization provides a potential solution to this. For summarizing student feedback, first, the opinion targets should be identified and extracted. In this context, opinion targets such as “lecture slides”, “teaching style” are the important key points in the feedback that the students have shown their sentiment towards. In this paper, we focus on the opinion target extraction task of general student feedback. We model this problem as an information extraction task and extract opinion targets using a Conditional Random Fields (CRF) classifier. Our results show that this classifier outperforms the state-of-the-art techniques for student feedback summarization. Keywords: Student Feedback Summarization; Opinion target Extraction, Conditional Random Fields 1. Introduction Student feedback is used widely in present in order to enhance the quality of teaching and learning. Feedback is collected from students as online forms as well as handwritten documents. Since it takes a considerable effort to read and understand all the feedback given by the students, the best way is to read all the feedback and create a summary that covers all the aspects of all the feedback given. 295 The 2017 Conference on Computational Linguistics and Speech Processing ROCLING 2017, pp. 295-307  The Association for Computational Linguistics and Chinese Language Processing Although many lecturers collect student feedback, comments written by students are not summarized. If a lecturer wants to get a summary of these comments, the lecturer has to manually read and summarize these comments. Manual summarization is not scalable; in a large class with more than few hundred students, it is going to be a tedious and rigorous task. Thus, a system to summarize all student feedback and giving an overall summary by categorizing students’ sentiments towards different aspects of the lecture will be very useful for teachers, lecturers, schools, universities, and the education systems as a whole. Research done in this area so far has focused only on using student feedback collected using reflective prompts [1]. With a reflective prompt, student feedback is collected by giving a specific question (prompt). For an example, a prompt such as “What are the most interesting topics of today’s lecture?” is considered as a reflective prompt. In reflective prompts, the prompt decides the opinion of the feedback: positive or negative. Opinion for different aspects in student feedback cannot be measured in this approach. In this paper, we focus on general student feedback. General feedback means that the feedback is collected using a general prompt (example: “Give feedback on today’s lecture”), rather than a specific prompt where the prompt suggests the sentiment of the feedback. Our system contains three parts: (1) Identifying and extracting all the opinion targets …

Content truncated to 500 words