Personality Evaluation Of Student Community Using Sentiment Analysis

The social media is growing very rapidly, especially in the youth within the student community. The student community or youth communicates through their postings in form of textual data, images and videos to convey their messages within their circles of closely known and unknown friendships. The postings posted by the youth may be collected and used for the research purposes. In the large ocean of postings and the posters it is difficult and necessary to identify the personality traits. This identification may be useful to know about the students who is evaluating the teacher in an academic institution, a product reviewer who is reviewing and commenting on the product, the client feedback toward the product or the company, a teacher who is in process to build the foundation of students and other so many areas. Facebook is a very famous and common platform where youth feels comfortable to share their views in form of comments in textual form. We had tried to develop a tool on python language using three classifiers MNNB, RF and SVC to predicting or identifying the personality of students on social media using four labels including "sad", "angry" (at negative side) and "happy", "relax" (at positive side). A design-based approach has been used based on a dataset extracted from social media self-created page(s) where students of different mind-set were invited to comment on the questions relevant to their academics e.g. "The role of good grades is more important to build the career of student instead of having good practical knowledge?" or "In the evaluation system the home assignments should be discouraged by encouraging the classroom activities and participation". In that case the dataset was collected fully a textual based having no emoticons used at all, the students expressed their views regarding the questions posted through social media page(s). Our approach succeeded to predict the personality with respect to posters Ids, time slots and shifts including morning, afternoon and evening using machine learning algorithms based on purely a textual dataset. In future the dataset and labels may be increased for more perfection in results to identify the personality, the personality may be predicted from the use of emoticons and roman English which is being used on social media very frequently.