Constructing a Feature-based Recommender System for One-to-One Marketing
碩士 === 輔仁大學 === 資訊管理學系 === 91 === In this paper, we construct a recommender system based on product features and the analysis of customer behaviors. We try to analyze why customers prefer some features of products from transaction databases and product databases, and then recommend the products that...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2003
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/86095113384255615154 |
| Summary: | 碩士 === 輔仁大學 === 資訊管理學系 === 91 === In this paper, we construct a recommender system based on product features and the analysis of customer behaviors. We try to analyze why customers prefer some features of products from transaction databases and product databases, and then recommend the products that may attract customers. The advantage of this paper is to solve the disadvantages of market basket analysis and collaborative filtering that are not able to recommend products which are not purchased previously or only purchased a few times. This research also makes use of a two-phased clustering technique that combines advantages of the Self-Organized Map and K-means Clustering techniques to find the neighbors of the target customers efficiently and recommend products based on their potential interests. Through mining customers’ preferences, we can find the reasons of the recommendations. Besides, it can help enterprises to develop new products based on customers’ preferences and use one-to-one marketing to increase profits.
|
|---|