面向情境化推荐服务的图书馆用户画像研究

Research on the Situational Recommendation-Oriented Library User Profiles

用户画像可以较为深刻地描述读者用户的群体行为特征,为图书馆的情境化推荐服务提供参考。基于图书馆用户的基本信息、内容偏好数据、互动数据、会话数据、情境数据进行了标签化识别,围绕这五个维度构建了图书馆用户画像,由此引入情境化推荐方法进行协同过滤推荐。研究结果表明,画像模型更好地将图书馆用户多维标签融入到情境化推荐过程中,缓解了传统协同过滤面临的推荐精度下降问题,为提升图书馆的个性化服务水平提供参考。

User profile is a tagged portrait abstracted by deeply mining of user demographic information, consumer behavior, social relations and situational information. This paper labels and identifies library users based on basic information, content preference data, interactive data, conversational data and situational data, constructs library user portraits, and introduces situational recommendation method. The results show that the proposed model integrates multi-dimensional labels of library users into situational recommendation process better, alleviates the decline of recommendation accuracy faced by traditional collaborative filtering methods, and provides a reference for improving library personalized service level.