基于模糊粗糙神经网络的用户个性化关联推荐方法  

The User Association Recommendation Method Based on Fuzzy Rough Neural Network

作　　者：陈翔[1] 唐俊勇[2] CHEN Xiang;TANG Junyong(School of Civil Engineering, Xi＇an Technological University;School of Computer Science and Engineering, Xi＇ an Technological University, Xi＇ an710021, China)

机构地区：[1]西安工业大学建筑工程学院 [2]西安工业大学计算机科学与工程学院,西安710021

出　　处：《重庆师范大学学报（自然科学版）》2018年第3期137-142,共6页Journal of Chongqing Normal University:Natural Science

基　　金：陕西省科技厅工业科技攻关项目(No.2016GY-088)

摘　　要：【目的】针对电子商务中关联推荐的有效性,基于模糊粗糙神经网络提出了一种新的方法。【方法】该方法首先提出了个性化推荐模型,并根据关联规则给出了推荐算法流程。同时利用模糊粗糙神经网络对上述模型进行求解,通过计算模糊关联系数和模糊关联度来获得用户行为特征。【结果】利用仿真实验深入研究影响该方法的关键因素,结果发现:最小可信阈值和最小支持度取值越小,预测精度就越高。【结论】与其他算法相比,该算法能在一定的程度上较小预测的误差,使得推荐结果更加满足客户的需求。[Purposes]In order to improve the effectiveness of the association recommendation in e-commerce,a new method is proposed with fuzzy rough neural network.[Methods]At first,the personalization recommendation model is presented in this model,and the recommendation algorithm process is given by the association rule. Then,it is solved with fuzzy rough neural network,and user behavioral characteristics are calculated with fuzzy correlation coefficient and correlation degree.[Findings]Finally,a simulation is conducted to study the key influence factor of this method. The result was found that the smaller the minimum trustworthiness threshold and the minimum support value,the higher the prediction accuracy.[Conclusions]Compared with other algorithms,this algorithm can reduce the prediction error to a certain extent,and make the recommendation result more satisfying to the customer＇s needs.

关 键 词：个性化 关联推荐 行为特征 模糊粗糙神经网络 预测精度 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]