| Summary: | 碩士 === 國立中央大學 === 資訊管理學系 === 103 === Recommended system is a common information filtering system. It’s a very important technology no matter in business or person, and it can provide users different documents according to their different interests. We can use ontology to conceptualize the concepts within the same domain. This study automatically construct ontology to let the unclassified document hierarchical clustering and applied to create user profile. Users’ interests can be mapped to ontology in order to manage interests by hierarchical clustering. Besides, this study improve the method to mapping user’s interests to ontology and added long term interest into the method to make the detection of users’ interests more accurate. In the recommendation, this study use structure similarity at the ontology to find the implicit interests and let the recommendation of documents be more diverse.
In the experiments, this study use book’s description on Amazon online shopping websites as data collection and simulate the change of users’ interest in different conditions. We could find that this study can improve the accurate of recommendation and make recommendation more diverse. Expects this system to be used for commercial, let enterprises to more accurately know the customer's interest and bring more benefits.
|
|---|
