摘要
[目的/意义]以丁香园社区中用户数据的多粒度知识建模为目标,提出基于模糊形式概念分析(Fuzzy Formal Concept Analysis,FFCA)的群体用户画像知识发现方法,揭示差异化群体用户与多维属性特征间的模糊关系,为实现粒度视角下用户需求的精准定位提供数据支撑。[方法/过程]通过数据预处理,抽取丁香园社区不同级别的站友数据,分别面向用户自然属性维、用户兴趣维、用户行为维建立群体用户细分标签组,实现群体用户画像的概念建模;通过引入用户兴趣模糊隶属度,建立用户细分模糊形式背景,并构建用户细分模糊概念格,诱导出模糊关联规则,完成群体用户画像的多粒度刻画。[结果/结论]在真实数据集上验证了该方法的可行性,实验表明将模糊概念格引入用户画像研究,利用偏序关系与模糊关系划分论域知识,有助于精准定位用户需求,弥补社会化标签不足的现实问题。
[Purpose/significance] In this work,the portrait of group users was constructed based on Fuzzy Formal Concept Analysis to reveal the fuzzy relationship between differentiated group users and multidimensional attribute features,which aims at multi-granularity knowledge modeling of user data.[Method/process] The model first extracted the data at different levels in Ding Xiang Yuan community.Afterwards,label groups of users could be established including three dimensions of the user’s natural attributes,user interests,and user behaviors to realize the conceptual modeling of group user portraits.The fuzzy formal context for user segmentation is established,by introducing the user fuzzy membership degree,depending on which fuzzy concept lattice for group user segmentation was constructed to induce fuzzy association rules,thereby making it relatively easy to obtain the multi-grained characterization of group user portraits.[Result/conclusion] The feasibility of the method was verified on a real data set.Experiments showed that the introduction of fuzzy concept lattices into user portrait studies and the use of partial order and fuzzy relationships could help to compensate for the lack of social labels.
出处
《情报理论与实践》
CSSCI
北大核心
2020年第8期103-111,共9页
Information Studies:Theory & Application
基金
安徽省高校人文社会科学重点项目“大数据背景下医疗纠纷事件的语义识别及其对网络舆情预警影响的研究”(项目编号:SK2018A1064)和安徽省高校人文社会科学重点项目“医院负面口碑补救行为对患者满意度影响与多维度关联分析”(项目编号:SK2018A1072)的成果。
关键词
用户画像
模糊概念格
多粒度
知识发现
模糊隶属度
user personas
fuzzy concept lattice
multi-granularity
knowledge discovery
fuzzy membership
