2018/05/08 by Abien Fred Agarap, Agarap, Abien Fred · 2 citations
Computer Science · #Sentiment Analysis and Opinion Mining
paper · pdf · doi:10.48550/arxiv.1805.03687
Understanding customer sentiments is of paramount importance in marketing\nstrategies today. Not only will it give companies an insight as to how\ncustomers perceive their products and/or services, but it will also give them\nan idea on how to improve their offers. This paper attempts to understand the\ncorrelation of different variables in customer reviews on a women clothing≠-commerce, and to classify each review whether it recommends the reviewed\nproduct or not and whether it consists of positive, negative, or neutral\nsentiment. To achieve these goals, we employed univariate and multivariate\nanalyses on dataset features except for review titles and review texts, and we\nimplemented a bidirectional recurrent neural network (RNN) with long-short term\nmemory unit (LSTM) for recommendation and sentiment classification. Results\nhave shown that a recommendation is a strong indicator of a positive sentiment\nscore, and vice-versa. On the other hand, ratings in product reviews are fuzzy\nindicators of sentiment scores. We also found out that the bidirectional LSTM\nwas able to reach an F1-score of 0.88 for recommendation classification, and\n0.93 for sentiment classification.\n