A Systematic Methodology Based on Word Embedding for Identifying the Relation Between Online Customer Reviews and Sales Rank

Abstract

In the buying decision process, online reviews become an important source of information. They become the basis of evaluating alternatives before making purchase decision. This paper proposes a methodology to reveal one of the hidden alternative evaluation processes by identifying the relation between the observable online customer reviews and sales rank. This methodology applies a combined approach of word embedding (word2vec) and X-means clustering, which produces product-feature words. It is followed by identifying sentiment words and their intensity, determining connection of words from dependency tree, and finally relating variables from the reviews to the sales rank of a product by a regression model. The methodology is applied to two data sets of wearable technology and laptop products. As implied by the high predicted R-squared values, the models are generalizable into new data sets. Among the interesting findings are the statements of problems or issues of a product are related to better sales rank, and many product features that are mentioned in the review title are significantly related to sales rank. For product designers, the significant variables in the regression models suggest the possible product features to be improved.

Analysis of Customer Sentiment on Product Features After the Outbreak of Coronavirus Disease (Covid-19) Based on Online Reviews

“…The feature extraction is composed of three steps: (i) preprocessing, (ii) phrase embedding, and (iii) phrase clustering. The methodologies used in this study are based on previous relevant frameworks and studies focusing on feature extraction and customer sentiment analysis (Suryadi & Kim, 2018;Park & Kim, 2020). To demonstrate the methodology, a case study is conducted using new and refurbished smartphone reviews to investigate the dynamic changes in customer sentiment during COVID-19 through online review analysis.…”

“…The result provided product features that caught the customer's attention. Suryadi & Kim (2018) identified the importance of product features on the sales rank by analyzing the relationship between each customer's sentiment score on features and the sales rank of the purchased product. However, to the best of our knowledge, there are few papers related to the dynamic changes in customers' sentiments about the features of products in connection with massive-scale disruption in the market such as COVID-19.…”

“…And then, various indexes have been used for detecting sentiments: (i) the sentiment score of the adjective connected to the target word; (ii) the sentiment level of the ICED21 sentences containing the target word; (iii) the average rating of the reviews mentioning the target word. This study follows the methodology of Suryadi & Kim (2018) using the first approach. First, each review sentence was scanned to check whether the sentence contains feature words.…”

“…To tackle this issue, this study aims to investigate the dynamic change of customer sentiment on product characteristics following COVID-19 through sentiment analysis. Sentiment analysis is one of the main fields of natural language processing (NLP) and aims to determine the attitude of a person concerning some topic or the overall contextual sentiment polarity of a document (Suryadi & Kim, 2018;Turney & Littman, 2003). Given that online purchases have increased during COVID-19 (Krakowski, 2020) and quick analysis and response in this crisis are needed, analysis of online reviews during COVID-19 has an advantage of being able to detect dynamic changes in customers' preferences for a short period compared to traditional methods such as surveys and interviews.…”

Research perspectives in ecodesign

“…The authors claim that compared to Super 4-Points Congruent Sets (S4PCS), their algorithm greatly reduces computational complexity by incorporating volumetric information from 3D models. Suryadi & Kim (2018) investigated the relations between online customer reviews and sales rank, aiming to identify the design features significantly related to sales rank. Their proposed methodology essentially consists of the following stages: (1) produce product-feature words by employing a combined word embedding-X-means clustering approach; (2) identify sentiment words and their intensity, and connect feature with sentiment words using a dependency tree;…”