中文摘要：

目的：为了使用户兴趣模型更好地被推荐系统理解，实现个性化服务。方法：引入领域本体，对特征项语义进行扩展，构建电影领域本体。根据信息论思想，改进相似度的计算方法，构建基于本体的用户四元组多兴趣细粒度表示模型和相应更新机制。结果：随着用户欣赏同类电影的数量的不断增加，用户兴趣模型能进行累加学习，推荐的准确性不断提高。当用户的兴趣爱好发生转移时，用户兴趣模型能随着用户兴趣的转移合理地“遗忘”掉用户过去的爱好，而积累用户新近感兴趣的电影主题。结论：实验表明该用户兴趣模型能够准确及时地跟踪用户多种兴趣及其变化，保证用户模型的可靠性。

英文摘要：

Purpose： Recommend system can understand the user model so ao to realize personalized service. Methods： We used the domain ontology to extend the semantic features of items and built the film domain ontology. According to information theory, we improved the similarity calculation method. We built the user interest model which make up of the four elements and base of ontology, and designed the corresponding update mechanism. Result： When the Users enjoy the more same movies, the user model constantly learned the knowledge, so the recommend system＇s accuracy continues to increase. When the user＇s interests turn to change, the user model reasonably forgotten the user＇s hobby in the past and accumulated the new interest for new movie. Conclusions： Experiments show that the user model can accurately track a variety of user＇s interests and its changes to ensure the reliability of the user model.