基于模糊逻辑的计算机网络课程资源个性化推荐方法  

Personalized recommendation method of computer network course resources based on fuzzy logic

作　　者：王尧[1] 蔡秋茹 于志敏[1] 罗烨[1] WANG Yao;CAI Qiuru;YU Zhimin;LUO Ye(Jiangsu University of Technology,Changzhou 213000,China)

机构地区：[1]江苏理工学院,江苏常州213000

出　　处：《无线互联科技》2025年第6期125-128,共4页Wireless Internet Science and Technology

基　　金：2024年江苏省高校“新时代教材数字化建设研究”专项课题,项目名称:面向新工科教材建设数字化实现的途径研究——以计算机网络课程为例,项目编号:2024JCSZ11;江苏理工学院教学改革与研究项目,项目名称:OBE视阈下“互联网+教育”教学新模式的实践探索——以计算机网络课程为例,项目编号:11610312311;江苏理工学院教学改革与研究项目,项目名称:OBE视域下应用型本科高校产教融合实施路径研究,项目编号:11610312314。

摘　　要：由于传统方法在提取用户行为特征时,直接采用原始数据而未计算分词的相似度,导致用户意图和偏好刻画存在偏差,使得特征提取存在稀疏性,进而导致推荐的准确性较低。为此,文章提出基于模糊逻辑的计算机网络课程资源个性化推荐方法。该方法通过融合多维度行为数据计算行为数据综合指数,构建加权行为数据矩阵并定义用户兴趣特征向量;再通过模糊隶属度函数将特征向量赋予隶属度值,从而构建用户兴趣模糊模型;利用模型得出匹配资源需求,评估特征相关性并计算特征兴趣度;根据特征重要性计算资源整体兴趣度,引入调整因子和自适应特征量调整权重并重新计算兴趣度,最后根据筛选阈值筛选资源,经多次排序实现计算机网络课程资源的个性化推荐。实验结果表明,该方法的归一化折损累积增益值更平稳且增长趋势较佳、命中率更高。这些结果充分证明了该方法在处理复杂推荐任务时的稳定性和准确性。Because the traditional method directly uses the original data without calculating the similarity of word segmentation when extracting user behavior features,it leads to the deviation of user’s intention and preference description,which makes the feature extraction sparse and leads to the low accuracy of recommendation.Therefore,the article proposes a personalized recommendation method of computer network course resources based on fuzzy logic.This method calculates the comprehensive index of behavior data by fusing multi-dimensional behavior data,constructs a weighted behavior data matrix and defines the user interest feature vector.Then,the feature vector is given the membership value through the fuzzy membership function,so as to construct the fuzzy model of user interest;Using the model,the matching resource requirements are obtained,the feature correlation is evaluated and the feature interest is calculated.According to the importance of features,the overall interest degree of resources is calculated,the adjustment factor and adaptive feature quantity are introduced to adjust the weight and recalculate the interest degree.Finally,the resources are screened according to the screening threshold,and the personalized recommendation of computer network course resources is realized through multiple sorting.The experimental results show that the normalized cumulative gain value of this method is more stable and the growth trend is better,and the hit rate is higher.These results fully prove the stability and accuracy of this method in dealing with complex recommendation tasks.

关 键 词：模糊逻辑 计算机网络课程资源 用户兴趣模型 自适应特征量 个性化推荐列表 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]