Aspect-Based Opinion Mining on Student’s Feedback for Faculty Teaching Performance Evaluation

Irum Sindhu,Maheen Bakhtyar,Sher Muhammad Daudpota,Junaid Baber,Kamal Badar,Mohammad Nurunnabi

doi:10.1109/access.2019.2928872

Abstract

Students' feedback is crucial for academic institutions in order to evaluate faculty performance. Handling the qualitative opinions of students efficiently while automatic report generation is a challenging task. Indeed, most organizations deal with quantitative feedback effectively, whereas qualitative feedback is either processed manually or ignored altogether. This study proposes a supervised aspect based opinion mining system based on two-layered LSTM model. The first layer predicts the aspects described within the feedback and later specifies the orientation (positive, negative, and neutral) of those predicted aspects. The model was tested on a manually tagged data set constructed from the last five years students' comments from Sukkur IBA University as well as on a standard SemEval-2014 data set. Unlike many other LSTM models proposed for other domains, the proposed model is quite simple in terms of architecture which results in less complexity. The system attains a good accuracy using the domain embedding layer in both tasks: aspect extraction (91%) and sentiment polarity detection (93%). To the best of our knowledge, this study is a first attempt that uses deep learning approach for performing aspect based sentiment analysis on students' feedback for evaluating faculty teaching performance.

Highlights

In this era of digital world, mining people’s opinion is very crucial in order to dig out the useful insights and their sentiments regarding any specific entity
Aspect sentiment classification In this study, we have explored two essential tasks of aspect based sentiment analysis (ABSA) [7] i.e. Aspect Extraction (AE) and Aspect Sentiment Classification (ASC) for the reason that entity and time of the opinion is already known, and opinion holder is needed to be kept anonymous in the academic domain
METHODOLOGY we discuss the methodology of our proposed model in detail: we first demonstrate the creation of academic domain data, followed by preprocessing step; afterward, we describe skip gram model for generating domain word embedding; lastly, we explain the working mechanism of our two layer LSTM neural network for aspect extraction and aspect sentiment classification

Summary

Introduction

In this era of digital world, mining people’s opinion is very crucial in order to dig out the useful insights and their sentiments regarding any specific entity. It is a common practice that when it comes to decision making, individuals or organizations prefer to seek out others’ opinions [1]. In academia, faculty teaching performance is evaluated through student’s feedback provided at the end of each course. The quantitative responses are aggregated and used as a measure to gauge the teaching quality of concerned faculty members. This is considered as one of the key factors in the annual appraisal process. Though student feedback form is comprised of closed ended questions, it provides students with a space for textual comments, an open ended feedback to express their thoughts

Objectives

Methods

Results

Conclusion