Analysing consumer comments on e-commerce websites using text mining methods

Today, as a result of rapid developments in technology, consumers prefer the online shopping environment more than the traditional shopping environment. Consumer comments are one of the most essential resources that customers benefit from when purchasing a product online. The comments shared by the customers who bought the product affect the purchasing decision of other customers and their thoughts about the brand of the product. However, there are numerous needless remarks unrelated to the product in the product comments, making it difficult to find the comments that provide information on the goods. On the other hand, in the comments shared by consumers, in cases where the words are written in spoken language, shortened, or excessive use of punctuation marks, the analysis of the text data to be processed becomes difficult and text mining techniques are used to overcome these difficulties. Therefore, the purpose of this study is to figure out, using text mining techniques, if online comments posted by customers who have previously made purchases from online shopping sites are relevant to the product. For this purpose, at first, a data set of 4795 comments on 21 products from various categories was constructed on www.trendyol.com, one of Turkey's top online marketplaces. Then, word vectors were created on this data set by using Term Frequency and Inverse Document Frequency methods. Finally, the comments were classified using the k Nearest Neighbor, Naive Bayes, Decision Trees, Support Vector Machine (SVM) and Deep Learning classification algorithms, and the highest success was obtained with the SVM algorithm with an accuracy value of 81.61%. It is thought that this study will make it easier for consumers to decide at the purchasing stage, as it will provide the opportunity to filter only the comments about the product. Additionally, online retailers will have the chance to assess their flaws and shortcomings by considering the comments made by customers regarding the merchandise.