Sentiment Analysis and Effective Visualization of Faculty and Course Feedback

Abstract

One of the most important ways to ensure the success of any product is to understand the mindset of the users/consumers. In order to achieve that understanding, one has to collect their opinion in the form of feedback. Hence feedback collection serves as an important goal towards the promotion of the product. This can be analogous to any educational body like a School or a University. The standard of the education can be improved greatly through the feedback data obtained from the students. But the key point here is to correctly interpret the opinions collected. Sentiment analysis techniques, as the name implies aids in understanding the sentiments of the users/students. These techniques determine whether the sentiment has a penchant towards positivity or negativity majorly. Although there is the concept of neutrality, this paper aims on categorizing the comments as positive and negative. The main aim of this paper to analyze the feedback data collected to come up with results that focus on the highlighting how much the students are learning and improving themselves and how much the faculty and the courses offered are helping them achieve that. Students' opinions on both faculty and course are collected in the form of comments through an online form. The data collected is then subjected to Cleaning and Tagging. The parameters that are taken into account for the faculty are communication skills, knowledge base, motivating factor and practical implementation of concepts. There is also an extra column for students to enter other opinions if any. Similarly, for the course feedback the parameters considered are relevance of course, course coverage, availability of learning material and scope for learning new concepts. The opinions collected are then assigned a specific value called Semantic Orientation (SO) based on the polarity of the comment, i.e. whether the comment falls under positive or negative category. Analysis is then performed based on the SO value to get an idea about the polarity of the comment. The values are then plotted as analytical graphs as an attempt to provide better visualization of the obtained results. The proposed system does better analysis at that it does not rely totally on scaling to obtain the results.