引进一种新型高准确度基于资源分配协同推荐方法,利用物质分配过程计算用户相关性。在此基础上考虑类型对相关性影响对算法进行了改进,增加系数λ调节类型因素影响相关系数程度。改进算法平均排名分数减小、平均度减小和平均Hamming距离增加,证明了推荐算法的准确性、多样性、个性化,并且算法时间复杂度也比标准协同算法明显减小。
Based on mass diffusion,this paper introduced a higher accuracy collaborative filtering(MDCF).By using the opinion diffusing process,the similarity between any users could be obtained.Furthermore,by consider the type correlations,added a tunable parameter λ change the effect of the type correlations.The smaller average ranking score,the lower mean degree and the bigger hamming distance of the improved algorithm show the algorithm results more accuracy,diversity and popularity.Furthermore the algorithmic complexity less than the standard collaboration.