CAJ | 학술논문

协同过滤推荐算法是个性化推荐服务系统的关键技术,由于项目空间上用户评分数据的极端稀疏性,传统推荐系统中的用户相似度量算法开销较大并且无法保证项目推荐精度.通过对共同感兴趣的项目属性的相似用户进行聚类,构建了不同项目评价的用户相似性,设计了一种优化的协同过滤推荐算法.实验结果表明,该算法能够有效避免由于数据稀疏性带来的弊端,提高了系统的推荐质量.
협동과려추천산법시개성화추천복무계통적관건기술,유우항목공간상용호평분수거적겁단희소성,전통추천계통중적용호상사도량산법개소교대병차무법보증항목추천정도.통과대공동감흥취적항목속성적상사용호진행취류,구건료불동항목평개적용호상사성,설계료일충우화적협동과려추천산법.실험결과표명,해산법능구유효피면유우수거희소성대래적폐단,제고료계통적추천질량.
Collaborative filtering recommendation algorithm is key technologies of personalized recommendation system, as the serious sparsity data of rated items, the similar users of active user is distribution of scattered. The traditional collaborative filtering recommender system algorithm consumes too many resources to search the nearest neighbor, as well as reliability is poor. A novel collaborative filtering recommender algorithm based on user clustering of item attributes is proposed. This algorithm reduces the negative effect on quality of recommendation, and shrinks the searching scope, which improves recommendation results for the system.