首页|基于LDA主题模型的MOOC推荐方法研究

基于LDA主题模型的MOOC推荐方法研究

扫码查看
随着MOOC平台数量的不断增多,课程资源的井喷式增长,用户很难在较短时间内快速找到自己感兴趣的课程.针对上述问题使用基于LDA分类降维的协同过滤算法为用户提供课程个性化推荐.首先,应用吉布斯采样法来估计LDA主题模型的多项式分布,对MOOC课程文本信息进行聚类.其次,根据课程信息聚类结果结合用户行为集合构建类别—评分矩阵.最后,根据类别—评分矩阵信息,计算用户余弦相似度,对目标用户进行协同过滤推荐.实验结果表明,基于LDA分类降维的协同过滤算法的准确率、召回率、综合评价指标F1值以及预测评分准确度对比传统推荐算法都有所提升.
Research on MOOC Recommendation Method Based on LDA Topic Model
With the increasing number of MOOC platforms and the explosive growth of course resources,it is difficult for us-ers to quickly find the courses they are interested in a relatively short period of time.Aiming at the above problems,a collaborative filtering algorithm based on LDA classification and dimension reduction is used to provide users with personalized course recommen-dations.First,the multinomial distribution of the LDA topic model is estimated by using Gibbs sampling method to cluster the MOOC course text information.Secondly,a category-scoring matrix is constructed according to the clustering results of the course information combined with the user behavior set.Finally,according to the category-rating matrix information,the user cosine simi-larity is calculated,and the target user is recommended for collaborative filtering.The experimental results show that the accuracy,recall rate,comprehensive evaluation index F1 value and prediction score accuracy of the collaborative filtering algorithm based on LDA classification and dimension reduction are improved compared with the traditional recommendation algorithm.

MOOC platformpersonalized recommendation of coursesLDA topic modelclusteringcollaborative filtering

杜梦晗、崔仙姬、姜雨蒙、张俊星

展开 >

大连民族大学信息与通信工程学院 大连 116600

MOOC平台 课程个性化推荐 LDA主题模型 聚类 协同过滤

2024

计算机与数字工程
中国船舶重工集团公司第七0九研究所

计算机与数字工程

CSTPCD
影响因子:0.355
ISSN:1672-9722
年,卷(期):2024.52(12)
  • 1