基于Apriori关联分析的协同过滤改进算法  

Improved Collaborative Filtering Algorithm Based on Apriori Association Analysis

作　　者：王琦 王逊 黄树成 WANG Qi;WANG Xun;HUANG Shucheng(School of Computer Science and Engineering,Jiangsu University of Science and Technology,Zhenjiang 212000)

机构地区：[1]江苏科技大学计算机学院,镇江212000

出　　处：《计算机与数字工程》2025年第3期617-622,665,共7页Computer & Digital Engineering

基　　金：国家自然科学基金项目“基于鲁棒表现建模的目标跟踪方法研究”(编号:61772244)资助。

摘　　要：在个性化推荐系统中,协同过滤技术被广泛采用。然而,其在处理数据稀疏性问题时的局限性,往往导致推荐结果的精准度不足。论文提出一种引入关联规则进行关联分析的协同过滤改进算法,首先,通过Apriori关联分析得到有效的强关联规则构建推荐度计算方法,并基于此进行评分预测。同时,考虑到传统的协同过滤推荐算法在推荐系统中的时效性问题,引入惩罚项和时间因子对原有的相似性度量算法进行优化,降低相似度计算时的偶然性对结果造成的误差。最后,通过集成推荐策略将两种不同的推荐算法融合以生成混合推荐结果。实验结果表明,与经典的协同过滤方法相比,所提出的改进算法在缓解数据稀疏性问题和提升预测准确性方面表现出显著优势,从而增强了推荐系统的性能。Collaborative filtering techniques are widely used in personalized recommendation systems.However,their limitations in handling data sparsity often lead to insufficient accuracy in recommendation results.This paper proposes an improved collaborative filtering algorithm by introducing association rule mining for association analysis.Firstly,effective strong association rules are obtained through Apriori association analysis to construct a recommendation score calculation method,which is then used for rating prediction.At the same time,considering the timeliness issues of traditional collaborative filtering algorithms in recommendation systems,penalty terms and time factors are introduced to optimize the original similarity measurement algorithm,reducing the error caused by the randomness of similarity calculations.Finally,a hybrid recommendation result is generated by integrating two different recommendation strategies.Experimental results show that,compared with classical collaborative filtering methods,the proposed improved algorithm significantly alleviates the data sparsity problem and enhances prediction accuracy,thereby improving the performance of the recommendation system.

关 键 词：协同过滤 关联分析 时效性 相似度 混合推荐 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]