LAUSR

These days users are able to save their time and effort by purchasing products online via various e-commerce websites. Their experience with a product exists in the form of textual reviews/feedbacks provided by them. Recommender systems offer personalized choices to users by capturing their interests and preferences. Through this paper identification of underlying topics using existing topic modeling techniques in user provided reviews of Moto e5 mobile on e-commerce website Amazon has been done and these techniques contrasted. Topic modeling is unsupervised learning technique used to identify hidden topics from a document (all the reviews of a product in this paper’s context). Coherence score, a measure of goodness of a topic reflecting the quality of human judgment compares these techniques. The higher the coherence score, the topic is more coherent. Experiments performed reveal that LDA technique performed better on the scrapped dataset.