混合粒子群遗传算法的协同过滤推荐模型  

Collaborative Filtering Recommendation Model Based on Hybrid Particle Swarm Optimization and Genetic Algorithm

作　　者：吴彦文[1] 王洁[1] 王飞[1] 

机构地区：[1]华中师范大学物理科学与技术学院,武汉430079

出　　处：《小型微型计算机系统》2017年第3期527-530,共4页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(71471073)资助;教育部重大支撑项目(2013BAH30F00)资助;湖北省高等学校省级教学研究项目(ccnu201315)资助;湖北省高等学校省级教学研究项目(ccnu201439)资助

摘　　要：在处理大数据方面,协同过滤算法在用户相似度计算方面存在数据稀疏性和推荐效果不明显的问题.利用协同过滤算法的优点,引入时间模型和信任度模型进一步过滤,提高协同过滤算法的精确率;同时采用粒子群算法和分层遗传算法进行混合优化,分层遗传算法可弥补粒子群算法容易陷入局部最优的缺陷,粒子群算法可加快混合算法的收敛速度;在Book Crossing数据集上做推荐对比实验.结果表明,在得到最近邻用户群的基础上,使用粒子群算法和分层遗传算法进行资源推荐,相比其他资源推荐算法,提高了资源推荐的质量.In dealing with big data,collaborative filtering algorithm has many problems such as data sparsity and the lack of obvious recommend effect in user similarity. The method integrated timing model and trust model to improve the precision of collaborative filtering algorithm by the advantage of collaborative filtering algorithm. Then PSO （ particle swarm optimization） and GA （ genetic algorithm ） were used to hybrid optimization, GA was applied to avoid falling in the local optimum, PSO was adapted to accelerate the convergence speed. After that , experiments were carried out on the BookCrossing dataset. The results show that,the algorithm which uses PSO combine hierarchical genetic algorithm for materials recommendation can improve the quality of materials recommendation by compared with the other materials recommendation algorithms based on the nearest neighbor users.

关 键 词：协同过滤 粒子群算法 分层遗传算法 混合优化 资源推荐 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]