| Summary: | Online customer reviews are valuable user-generated content to help in purchasing decision process. There are many customer reviews available in today's online marketplaces, allowing customers to evaluate them for a better knowledge of the product or service they will be purchasing. In addition to the recommendation system, expert opinions, and product descriptions, online customer reviews play a significant supporting role. However, due to the abundance of reviews, it is unclear whether these reviews are of high quality and useful to other customers. New online customer reviews may be of high quality, but because they are not widely known, other customers may overlook them. As a result, it is necessary to assess the quality of customer reviews that able the reviews to help a customer in their decision-making process. Existing studies tend to ignore quality indicators such as length and readability that provide a way to evaluate the important characteristic by employing a simple statistical analysis method. Hence, we propose the implementation of the Laplacian Score Algorithm to measure the readability index importance in the selection of the top-5 readability measure, to be used with other quality features in measuring the review quality. To validate the proposed framework, we use six review datasets from Amazon. The result shows that the basic structural information of a review text such as word, character, sentence and syllable counts give more influence to the quality of the reviews. Whereas advanced structural information such as difficult words and polysyllable count has less significance to the determination of the quality of a review text when it involves only the readability index measurement. © 2022 IEEE.
|
|---|