CAJ | 학술논문

针对实时推荐过程中实际数据的稀疏性，满足条件的项目或用户较少，导致推荐精度较低的问题，提出一种采用抽样近邻的协同过滤算法。该算法充分利用评分用户矩阵提供的信息，增加了参与到预测评分计算过程中的用户或项目，从而解决了传统协同过滤算法在实际应用中的不足。实验结果表明，在增加在线计算时间较少的情况下所给算法可有效提高推荐精度。
침대실시추천과정중실제수거적희소성，만족조건적항목혹용호교소，도치추천정도교저적문제，제출일충채용추양근린적협동과려산법。해산법충분이용평분용호구진제공적신식，증가료삼여도예측평분계산과정중적용호혹항목，종이해결료전통협동과려산법재실제응용중적불족。실험결과표명，재증가재선계산시간교소적정황하소급산법가유효제고추천정도。
Since the user-item matrix is sparse,and there are less users or items satisfying the conditions,the precision of the algorithm can’t be high.By sampling neighbor collaborative filtering algorithms,users take full advantage of score matrix provided information to increase the users or projects participated in the calculation process,so as to solve the shortage of traditional collaborative filtering algorithms in real application. Experiment results show that the new algorithm can effectively improve the precision in recommendation along a small increasing of runtime.