Extraction of Competitive Factors in a Competitor Analysis Using an Explainable Neural Network

Abstract

A competitor analysis is a core process in management decision making, and the extraction of competitive factors is a key component in a competitor analysis. Owing to the rapid development of social media, methodologies and frameworks facilitating a competitor analysis through online reviews have recently been proposed. However, existing studies have only focused on the detection of comparative sentences in review comments. Thus, product developers compare the competitive factors only within identical feature dimensions of a product. This study proposes a novel approach to identifying the competitive factors at the comprehensive level of the product features—in other words, the focus is on the points of differentiation of each product as perceived by customers using an explainable neural network. We assume that keywords, which have a significant influence on the classification decision, are considered meaningful points of differentiation by consumers. Thus, we establish a classification model to classify the review comments for each corresponding product and calculate the weight of importance of each keyword in such comments during the classification process. We then extract and prioritize the keywords to determine the competitiveness based on the weight of importance. The experiment results show that our proposed method effectively extracts the competitive factors in both qualitative and quantitative experiments. This is the first study on extracting competitive factors with an explainable neural network based on customer reviews; future studies may extend the scope of the extraction of such factors using an explainable neural network.