張鵬程

摘? 要： 傳統(tǒng)的遠程教育課程推薦方法因數(shù)據(jù)稀疏問題，造成其主題集中性較差，為此設(shè)計基于LDA用戶興趣模型的遠程教育課程推薦方法。通過遠程教育課程外在屬性包容度和內(nèi)在屬性質(zhì)量值，計算遠程教育課程的重要度，并以重要度為依據(jù)，利用LDA用戶興趣模型判斷用戶對主題的偏好度，確定主題與遠程教育課程的相似度系數(shù)，獲得用戶對遠程教育課程的興趣度，以此為基礎(chǔ)完成遠程教育課程的推薦。實驗結(jié)果表明：使用基于LDA模型的推薦方法向用戶推薦的課程有50%以上都是用戶需求的課程，而傳統(tǒng)的推薦方法只有不到20%，兩者相比，基于LDA模型的推薦方法的主題集中性更強，更適合應(yīng)用在遠程教育課程推薦中。

關(guān)鍵詞： 遠程教育; 課程推薦; LAD用戶興趣模型; 主題確定; 重要度計算; 偏好度判斷

中圖分類號： TN911?34; TP301? ? ? ? ? ? ? ? ? ? 文獻標識碼： A? ? ? ? ? ? ? ? ? ? ?文章編號： 1004?373X（2020）03?0173?04

Research on distance education course recommendation method

based on LDA user interest model

ZHANG Pengcheng

（Henan Radio & Television University， Zhengzhou 450008， China）

Abstract： The traditional distance education course recommendation method is poor in topic concentration due to data sparsity. Therefore， a distance education course recommendation method based on LDA （latent Dirichlet allocation） user interest model is designed. The importance degree of distance education courses is calculated according to the inclusiveness of external attributes and the quality value of internal attributes of distance education courses. The users′ preference to the subject is determined by LDA user interest model on the basis of importance degree to determine the similarity coefficients between subjects and distance education courses and obtain users′ interestingness for distance education courses. The distance education courses are recommended according to the obtained users′ interestingness. The experimental results show that more than 50% of the courses recommended to users by the LDA model based recommendation method are the courses required by users， while the correctness of the courses recommended to users by the traditional recommendation method is less than 20%. In comparison with the traditional recommendation method， the LDA model based recommendation method has better topic concentration and is more suitable for distance education course recommendation.

Keywords： distance education; course recommendation; LAD user interest model; subject determination; importance degree calculation; preference degree judgment