基于科研人员情境化主题偏好的科技文献协同推荐研究

Research on Collaborative Recommendation of Scientific and Technological Literature Based on Researchers’Contextual Topic Preference

[目的/意义]针对当前数字图书馆科技文献现有推荐方法中存在的语义缺失、情境缺失及潜在偏好挖掘不足等问题,提出基于科研人员情境化主题偏好的科技文献协同推荐方法。[方法/过程]首先基于情境感知技术识别科研人员情境信息,其次引入文本语义技术LDA主题模型挖掘科研人员的初始情境化主题偏好,继而根据科研人员情境的相似度与协同过滤思想扩展科研人员的情境化主题偏好,最后基于融合后的科研情境化偏好构建满足科研人员情境化需求的科技文献推荐列表。[结果/结论]实验结果表明,文章提出的基于科研人员情境化主题偏好的用户模型,能够较好地预测科研人员偏好,推荐效果更佳。

[Purpose/significance]In order to solve the problems of lack of context,lack of semantics and insufficient potential preference mining in the current recommendation methods of scientific and technological literature in digital libraries,this paper proposes a collaborative recommendation method of scientific and technological literature based on the contextual topic preference of researchers.[Method/process]Firstly,the contextual information of researchers is identified based on context-aware technology.Secondly,a textual semantic technology topic model is introduced to mine the researchers’initial contextualized topic preferences,and then researchers’contextual topic preferences were expanded according to the similarity of researchers’context and collaborative filtering.Finally,based on the merged scientific research contextual preference,a recommended list of scientific and technological literature is constructed to meet the contextual needs of researchers.[Result/conclusion]The experimental results show that the user model based on researchers’contextual topic preferences proposed in this paper can predict researchers’preferences better and has better recommendation effect.