CAJ | 학술논문

协同过滤技术被广泛应用于各种推荐系统当中.基于内存的协同过滤算法通过比较目标用户与其他用户的已有评分,为目标用户的未评分项目作出相应的预测.提出了一种新的基于内存的算法.根据项目的关键属性对它们进行分类,通过计算用户对各类项目的认知度,为目标用户选择相似用户并预测评分.通过MovieLens数据集的实验结果表明,该算法可以有效地解决包括数据稀疏性和新用户在内的一些协同过滤的基本问题,提供更高质量的推荐.
협동과려기술피엄범응용우각충추천계통당중.기우내존적협동과려산법통과비교목표용호여기타용호적이유평분,위목표용호적미평분항목작출상응적예측.제출료일충신적기우내존적산법.근거항목적관건속성대타문진행분류,통과계산용호대각류항목적인지도,위목표용호선택상사용호병예측평분.통과MovieLens수거집적실험결과표명,해산법가이유효지해결포괄수거희소성화신용호재내적일사협동과려적기본문제,제공경고질량적추천.
Collaborative filtering is widely used in many recommender systems.Memory-based algorithms make predictions on unrated items for the active user by comparing his individual ratings on rated items with the other users'.In this paper,a novel memory-based algorithm is proposed,in which the items are classified by their key attributes,similar users are selected and ratings are predicted for the active user after computing cognition degree.Experimental results on MovieLens dataset show that this algorithm outperforms the other traditional memory-based collaborative filtering algorithms.