The importance of digital media, which is advantageous for both consumers and sellers, is increasing day by day. E-commerce, which is the ideal way for companies to reach a larger audience in a much shorter time, has become the most attractive shopping method for consumers. As in every industry, the essential cornerstone of the e-commerce industry, where the demand is growing, and consumer complaints and requests are increasing day by day, is customer satisfaction.
"Customer Feedback Classifier" model is provided to e-commerce companies that want to understand their customers better, respond to their requests, reduce complaints, and, more importantly, take a leading place in the industry. The model can classify customer reviews (no matter how many rows data it has) in 7 head-categories in the fastest and effortless way.
It is now easy to learn which categories are most reviewed and which topics should be dwelled on. It is also possible to use this model together with the Turkish sentiment model and make a positive and negative distinction for the seven different categories mentioned earlier.
About Training Set
The "Customer Feedback Classifier" was trained with a customized dataset with related keywords from our Analytics platform. Dataset consists of 4600 customer reviews about different companies and brands in the e-commerce industry, and it contains seven categories: price, quality, customer service, mobile applications, sales promotions, advertisement, and delivery.
Before the last version of this dataset was reached, brand and company names were cleared from the data to prevent brand names from being associated with the categories and to be mislabeled. Unrelated comments and reviews from the specified categories were removed from the data.
The training dataset was tested with five different data consisting of reviews made to e-commerce companies. Each wrongly labeled row was exemplified until correct labeling was achieved.