A Study of Product Recommendation in Multiple Channels
|關鍵字:||產品推薦;協同式過濾推薦;多元通路;顧客分群;序列規則;Product Recommendation;Collaborative Filtering;Multi-channel;Customer Segmentation;Sequential Rule|
In this information era, customer's purchase behavior has changed a lot. In the past, customers could only buy products in the physical channel: store. Nowadays, customer-oriented companies provide many virtual channels for customers and allow them to choose the channel that they like or get used to buy products. Companies can also reduce the expenses of physical channels with virtual channels, e.g., TV channel or catalog channel. In addition, companies use recommendation systems to recommend suitable products to customers based on their interests and buying history, so as to improve customer’s degree of satisfaction. Many methods are used to implement recommendation systems, such as collaborative filtering and the sequential rule-based recommendation method. Each method has its own merit and drawback. Past research presented the hybrid method, which combined the advantages of many recommendation methods. The hybrid method can aslo overcome the drawback of individual recommendation method and improve the effect of the recommendation. However, previous research did not consider the factor of multiple channels. Customer's purchase behavior may differ in different channels. This work combines the factor of multiple channels with recommendation methods and verifies the effect of multiple channels. The experimental results show that the recommendation methods that combine the factor of multiple channels improve the quality of recommendation.