An Exploration of the Model for EC Recommender Systems
| Autor: | Sung-Lin Tsai, 蔡松霖 |
|---|---|
| Rok vydání: | 2013 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 101 In recent years, there are three most important recommender systems such as Content-based Filtering, CBF、Collaborative Filtering, CF and Knowledge-based Filtering, FBF. However, all has its advantages and limitations. When probing the source of the problems, we’ve found that e-commerce stores are always lack of sufficient information and have difficulty to speculate effectively. It’s prone to call it Cold start problem (CSP). The research tries to study products recommended system and use the questionnaire survey to understand online consumer behavior, and use data mining (DM) mode analysis, classify the products into "general products recommended "and" complementary products recommended. Then we systematically explore the questionnaire data, and analyze consumer behavior. On one hand, we have a schematic description of consumer behavior; on the other hand, the model description will help explain the information that causes insufficient current (cold start problem). In addition to considerations of consumer characteristics and behavior record, we can take care of consumers when they are at their first time to think about to buy a new product. The reason how they visit the online store and why they not purchased are because of their "Subjective Cognition”. By qualifying the favorite products for inference models and deriving from every problem that may occur, it will be a better recommendation system model. Also the "complementary products recommended system " can extend further possibility of consumption, any kind of problems that recommendation system can help to analyze and deduce. The computational model is how to calculate consumer preferences of products scores for "general products recommended" model problems that may occur, and further propose a model based on general products under recommendation, that is an extension of consumer recommendation, namely, "complementary products recommended " model. These will gradually solve the problems that may occur. The last instance of scenarios is used to examine two models which are reasonable. It's a Win-Win Game. Our model of complementary products recommended, can stimulate shopping and reduce invalid recommendation in the same time. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|