基于交互历史的三维家居个性化推荐与虚拟展示  

Personality-Oriented Recommendation and Virtual Display for 3D Home Design Based on Interaction History

作　　者：黄东晋[1] 唐鹏斌 王音[1] 丁友东[1] 

机构地区：[1]上海大学影视艺术技术学院,上海200072

出　　处：《应用科学学报》2015年第4期407-418,共12页Journal of Applied Sciences

基　　金：国家自然科学基金(No.61402278;No.61303093);上海市自然科学基金(No.14ZR1415800);上海市科技攻关项目基金(No.14511108400)资助

摘　　要：为了满足家居设计的个性化需求,将个性化推荐算法引入移动交互式三维家居个性化设计系统中,提出了基于交互历史的协同过滤推荐算法.首先,分析了用户的历史交互行为,构建了一个新的用户兴趣度量度模型,将用户的交互行为转换成用户兴趣度矩阵.然后,综合考虑了家具单品的时效性和资源关联特性,并将这两种特性引入协同过滤推荐算法的生成推荐过程中,以提高推荐质量.最后,将系统推荐的家具单品应用到三维虚拟家居设计场景中,通过对场景个性化编辑与虚拟展示,完成家居的个性化设计.实验结果表明,该方法是可行的,能有效提高推荐质量,具有较好的用户体验和视觉效果.To meet individual needs of home design, a collaborative filtering recommenda- tion algorithm based on interaction history is presented in this paper. The algorithm has been introduced into a mobile interactive 3D home design system. In the algorithm, user historical interaction behavior is analyzed, and a new user interest model built. The model converts user interaction behavior into a user interest matrix. To improve the recommen- dation quality, two furniture item characteristics, i.e., timeliness and resource relevance, are introduced into the generation process of the collaborative filtering recommendation algorithm. The recommended items are applied to 3D virtual home design scene. The home personalized design is completed by scene editing and virtual display. Experimental results show that the proposed method is feasible. It can effectively improve recommendation Quality with better user exoerience and visual effects.

关 键 词：虚拟家居 个性化推荐 用户兴趣度模型 协同过滤 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]