Finding the reviews on yelp that actually matter to me: Innovative approach of improving recommender systems |
| |
| Institution: | 1. National Engineering Research Center for Multimedia Software, School of Computer Science, Wuhan University, China;2. Institute of Artificial Intelligence, Wuhan University, China;3. School of Computer Science, Wuhan University, China;4. Xiaomi Inc., Wuhan, China;1. School of Information Science and Engineering, Shandong Normal University, No. 88 East Wenhua Road, Jinan 250014, China;2. School of Mathematical Science, Shandong Normal University, No. 88 East Wenhua Road, Jinan 250014, China;1. College of Intelligence and Computing, Tianjin University, Tianjin 300350, China;2. School of Physics, University of Science and Technology of China, Hefei 230026, China;3. Microsoft Research Asia, Beijing 100080, China;4. School of Precision Instruments & Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China;5. School of Marine Science and Technology, Tianjin University, Tianjin 300072, China;6. Center for Biosafety Research and Strategy, Tianjin University, Tianjin 300072, China |
| |
| Abstract: | As many readers struggle with massive textual information on review websites, developing optimized recommender systems that assist readers in identifying relevant reviews is critical. The present study aims to explore and predict the relationship between a reviewer’s evaluation of distinct attributes (i.e., importance and sentiment of a restaurant aspect)2 and overall satisfaction (i.e., generic numerical rating of a restaurant). Latent Aspect Rating Analysis is modified to achieve the goal. The study identifies five restaurant attributes: food & drinks, customer service, dining atmosphere, restaurant value, and location. Restaurant value contributes most from the importance perspective and food & drinks contributes most from the sentiment perspective. Restaurant value ranks the first as the overall satisfaction of attributes (i.e., combination of importance and sentiment). Accordingly, the present study suggests a supplement of the “dynamic” recommender systems. This study offers scholars and practitioners a refined approach to analyze wealthy review content. |
| |
| Keywords: | Recommender systems Yelp Latent aspect rating analysis (LARA) Natural language processing (NLP) Machine learning |
| 本文献已被 ScienceDirect 等数据库收录！ |
|
