协同显式和隐式主题回归的论文推荐算法  

Collaborative Explicit and Implicit Topic Regression Algorithm for Paper Recommended

作　　者：孔功胜[1] 

机构地区：[1]河南科技大学图书馆,河南洛阳471023

出　　处：《河南科技大学学报（自然科学版）》2017年第5期43-46,共4页Journal of Henan University of Science And Technology:Natural Science

基　　金：国家自然科学基金项目(61370220);河南省高校科技创新团队支持计划基金项目(15IRTSTHN010)

摘　　要：针对传统论文推荐模型存在冷启动和推荐解释困难的问题,在协同主题回归的基础上,建立了一种协同显式和隐式主题回归的论文推荐模型。模型中显式主题能够表示用户对论文内容中包含显式内容的偏好程度,隐式主题能够覆盖论文内容中没有显式包含的内容,该算法能够对推荐结果给出一定的解释。在CiteULike数据集上的实验结果表明:在不同推荐列表长度下,协同显式和隐式主题回归模型的推荐精度和召回率优于协同过滤和主题回归模型。Consider the problems of cold start and difficult interpretation in traditional paper recommended model,a new recommended model of collaborative explicit and implicit topics regression for papers recommended was represented based on collaborative topic regression. The explicit topics represented the preference of the users to the explicit content of papers,and the implicit topic covered features that were not included in the content of papers. The algorithm gave some interpretation to the recommended results. The experimental results on CiteULike data show in different length of recommend list,the model of collaborative explicit and implicit regressions are better than the collaborative filtering and topic regression model in recommended precision and recall rates.

关 键 词：推荐系统 协同过滤 主题回归 概率模型 论文推荐 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]