Abstract：Most of the current recommendation algorithms aim to improve the accuracy of the recommendation list,but less on the diversity of recommendations,which will result in less novelty of the recommendation results.And when considering diversity,it will reduce the accuracy.This paper proposes a diversified recommendation algorithm that combines interest distribution and singular value decomposition.Firstly,the kernel density estimation method is used to estimate the user interest distribution,and the neighbors with similar interest distributions are obtained,and the interest distribution is used.The similar neighbor's score pre-scores the items that the current user has not scored,and then fills the pre-score obtained in the previous step into the user-scoring matrix for SVD decomposition.At this time,the recommendation list includes the similarity neighbor pre-score.The items that come in are guaranteed the diversity of the recommended list,and there are also items that are similar to neighbors that are decomposed by SVD,ensuring the accuracy of the recommended list.Experiments on experimental data sets show that the proposed algorithm can improve the recommendation diversity under the condition of ensuring accuracy.
李卫疆,罗潘虎. 融合核密度估计和奇异值分解的多样化推荐算法[J]. 小型微型计算机系统, 2020, 41(1): 56-60.
LI Wei-jiang,LUO Pan-hu. Diversified Recommendation Algorithm Integrating Kernel Density Estimation and Singular Value Decomposition. Journal of Chinese Computer Systems, 2020, 41(1): 56-60.
2020, Vol. 41 
