社会化标注系统标签质量影响因素研究:基于随机森林算法

Research on Influencing Factors of Tag Quality in Social Tagging System: Based on Random Forest

[目的/意义]在社会化标注系统中,标签质量往往关乎用户对网络资源的分类、查询、浏览、获取等使用体验,确定影响标签质量的关键因素有助于进一步优化社会化标注系统的资源组织核心功能。[方法/过程]以社会化标注系统的标签为研究对象,从标注主体、标注客体、标注环境、标注动机、标注方式、标注产物等维度入手重构标签质量影响因素模型,尝试探究影响社会化标签质量的关键因素,并运用问卷调查方法收集数据,结合有监督学习的随机森林算法,建立标签质量影响因素的决策树模型。[结果/结论]结果显示,标注主体是影响标签质量的首要关键维度,主体的知识结构和认知水平、标注频度及其感知有用性对标签质量的影响突出;标注方式是影响标签质量的次要关键维度,标签推荐和规范标签提示是影响标签质量的重要因素。

[Purpose/significance]Tag quality is often related to users'experience of classification,query,browsing,acquisition of online resources in social tagging system.Identifying key influencing factors of tag quality can optimize the core functions of resources organization of STS.[Method/process]Based on tags,we provided the influencing factors model of tag quality from six perspectives,which covered tagging subject,tagging object,tagging environment,tagging motivation,tagging methods and tagging products.The study attempted to explore the key influencing factors of tag quality by questionnaire,and established the decision tree model of influencing factors of tag quality based on Random Forest.[Result/conclusion]Tagging subject is the primary key dimension affecting tag quality.And the impact of the subject's knowledge structure and cognitive level,the subject's tagging frequency,and the subject's perceived usefulness are prominent.Tagging methods are the secondary one,and tag recommendation and standard tag tips are main influencing factors.