Abstract
E-commerce platforms offer access to rich information about products, including customer-written reviews. This new data is used as signal for result list ranking, dynamic facets, or to generate product descriptions. In this paper, we put review data to the test and analyze their potential to serve as a reliable data source for information about products. For 50 products, we compared product details mentioned in reviews with the actual product information and identified several pitfalls, including customers talking about different products, reporting context-dependent values, and uttering about
desired product specifications. With this work, we highlight challenges that need to be accounted for when employing automatically extracting information from customer review in e-commerce.
desired product specifications. With this work, we highlight challenges that need to be accounted for when employing automatically extracting information from customer review in e-commerce.
Original language | English |
---|---|
Number of pages | 11 |
Publication status | Published - Jul 2024 |
Event | SIGIR Workshop on eCommerce, SIGIReCom 2024 - Washington, United States Duration: 18 Jul 2024 → 18 Jul 2024 |
Workshop
Workshop | SIGIR Workshop on eCommerce, SIGIReCom 2024 |
---|---|
Abbreviated title | SIGIReCom 2024 |
Country/Territory | United States |
City | Washington |
Period | 18/07/24 → 18/07/24 |
Keywords
- Information Extraction
- E-Commerce
- Review Data
- Product Description