Sorting through CULPA reviews for potential professors during the Add/Drop period adds an extra layer of stress to an already stressful time. Muddled with too-extreme reviews and not-enough reviews, CULPA often makes it hard for students to objectively determine whether a professor is gold-nugget-worthy or no-nugget-worthy. dietCULPA provides a simple interface for students to search up potential professors and get the most important information quickly.
A sample of 5000 reviews were parsed, word-by-word, to determine the average "sentiment value" based on a large .csv pairing some value between -1 (negative "sentiment") and 1 (positive "sentiment") created from analyzing Twitter feeds. All 30000 reviews on CULPA are normalized to this average via z-score. Upon querying the reviews of a professor, dietCULPA quickly determines exactly how many standard deviations away from his average professor colleague that a professor is favored by his students. With this data, we could also identify the most characteristic review of a professor, that is, that holds a "sentiment-value" closest to the average "sentiment-value" of the reviews of that professor.
But the true hallmark of dietCULPA is the .csv key file. Provided is a generic sentiments.csv file containing data pulled from Twitter. Modifying the key modifies exactly what kind of value is placed on each word found in a review. Any student can create one and thus can individualize the positions of professors on a distribution based on his own values. Say you value a professor who assigns a "low workload" - simply enter "low workload" and its synonymous phrases into the .csv key and give it a value of 1.0!
Log in or sign up for Devpost to join the conversation.