【作者】 杨萌；

【导师】 徐宝祥；

【作者基本信息】 吉林大学 ， 情报学， 2014， 博士

【摘要】 Web2.0下，用户对社会化标注系统资源的获取利用提出了多样化、专业化、个性化、人性化和知识化的需求，探索多维度的资源聚合已成为大势所趋与学科前沿。社会化标注系统资源聚合问题的求解是一个多值问题，而tax-folk混合模型就是其中一解。分析当前社会化标注系统资源聚合的现状，国内外的学者和专家们着重从四个方面展开研究：基于深度语义的社会化标注系统资源聚合、基于广度关联的社会化标注系统资源聚合、跨系统的社会化标注系统资源聚合和社会化标注系统资源导航，并取得了一定的成果。但遗憾的是，国内外的相关研究还仍有诸多不足，体现为尚未形成完善的社会化标注系统资源语义体系、缺乏对社会化标注系统资源聚合的机理剖析、社会化标注系统资源聚合的维度不完善、跨系统的资源聚合愈发迫切、社会化标注系统资源导航形式亟需多样化等等。社会化标注系统资源语义体系是社会化标注系统资源聚合的基础和重要支持。通过对比分析当前社会化标注系统资源语义表示形式的差异化和交互性，就可构建出具有“资源层-符号层-知识表示层-映射层-逻辑推理层”五层架构的社会化标注系统资源语义体系。有了该体系的支撑，就可在此基础上探讨资源多维度聚合的机理：各种语义表示形式应互动互补从而实现对资源深度语义的揭示，同时，以形式概念分析等工具实现基于各种数据集的概念分析，最终形成理论、方法和技术的有机融合。构建Tax-folk混合模型正是这一机理的具体体现。Tax-folk混合模型源于对taxonomy与folksonomy两种方法的融合。以理想的知识之树为比喻，其形态应为“干强枝繁叶茂”，而taxonomy形态为“干强枝繁叶少”，同时folksonomy的形态为“干少枝弱叶茂”，将folksonomy嫁接到taxonomy形成tax-folk混合分类才是解决社会化标注系统资源聚合的思路之一，这把嫁接之剪就是形式概念分析。同时，我们也必须看到，尽管tax-folk混合分类的思路是清晰的，但其实施涉及诸多要素的相互作用，这些要素包括组成要素、角色要素以及功能要素，这些要素和要素之间的深层关系共同组成了tax-folk混合模型，并从而从宏观层面回答tax-folk混合分类“做什么”的问题。进一步地，基于Tax-folk混合模型的社会化标注系统资源聚合流程则是从微观层面回答tax-folk混合分类“怎么做”的问题。通过数据准备、概念格构建、概念格分析、tax-folk映射、tax-folk混合分类树生成、输出与评价等六个关键步骤，就可以利用形式概念分析真正实现将folksonomy嫁接到taxonomy形成tax-folk混合分类的操作环节。以“豆瓣网”这一社会化标注系统为案例分析对象，以豆瓣读书中的“豆瓣五万至十万人读过”的图书资源为实验资源，收集相关数据并用基于tax-folk混合模型的社会化标注系统资源聚合流程进行相关操作，并对结果进行分析，最终证明该理论是科学、合理、可操作的。基于Tax-folk混合模型的社会化标注系统资源聚合研究，提出了社会化标注系统资源组织理想知识之树的新理念，探寻了构建tax-folk混合模型实现社会化标注系统资源聚合的新途径，形成的tax-folk混合分类树具有很高的资源组织优势，因而研究成果具有新价值。基于Tax-folk混合模型的社会化标注系统资源聚合研究的意义在于，一方面完善了社会化标注系统资源聚合理论体系，具有理论探索意义；另一方面，tax-folk混合分类方案有助于提高社会化标注系统的资源利用效率，进而带来经济效益和社会价值。更多还原

【Abstract】 Under the environment of Web2.0, users proposes more and more andneeds such as diversification need, professional need, personalized need,humane need and knowledgeable need to access the resources in socialtagging system，so it has become a trend and scientific frontier thatexploring multi-dimensional resource aggregation. We all know it is amulti-value solutions to solve the problem，and construction a tax-folkhybrid model is one solution.When we analyze the current situation of social tagging systemresource aggregation，we can find that scholars and experts from home andabroad has achieved some results and they often focus their research onfour areas：Resources aggregation based on depth semantic in socialtagging system, resources aggregation based on association in socialtagging system, resources aggregation across social tagging systems andresources navigation in social tagging system. Unfortunately, there arestill many unsolved issues to be explored, such as building an excellentsematic architecture is a new-emerging field in it, and the aggregationmechanism of social tagging systems is also required, and the resourceaggregation dimension of social tagging system is imperfect, also theresource aggregation across systems becomes more and more important, andfinally the form of resources navigation needs a diversification change.Resources semantic architecture is the basic and support of resourceaggregation of social tagging system. Through comparative analysis of thedifferentiation and interactivity of the resources semanticrepresentation in current social tagging systems, a new resourcessemantic architecture can be built, which consist of five layers: "resource layer-the symbol layer-knowledge representation layer-mapping layer-logical reasoning layer". With the support of thearchitecture, the aggregation mechanism of social tagging systems can beexplored: all of the semantic representation tools and methods shouldinteract and complete each other in order to achieve the goal of deeplyrevealing the resources’ semantic, while the concept analysis based ondata sets should be actualized using FCA et.al., and ultimately we canget an organic integration of theory, methods and techniques. Buildingthe Tax-folk hybrid model is just the concrete manifestation of thismechanism.Tax-folk hybrid model is derived from the integration of taxonomy andfolksonomy. In an ideal metaphor for the tree of knowledge, its shapeshould be "A tree with strong trunk, luxuriant branches and foliage",while the metaphor of taxonomy should be “A tree with strong trunk andbranches but little foliage”, and the metaphor of folksonomy should be“A tree with weak trunk and branches but luxuriant foliage”, so it isthe right solution of resource aggregation problem that we can graftfolksonomy to taxonomy in order to construct a new tax-folk hybridclassification, and the FCA tools are just like scissors.At the same time, we must also see that although the idea of tax-folkhybrid classification is clear, but its implementation involves manyinteraction factors, including the component elements, role elements andfunctional elements. These elements and the deep relationship among themtogether constitute a tax-folk hybrid model, which give a clear answerabout “what" should the new model do from the macro-level.Further, the specific process of resources aggregation based on theTax-folk hybrid model is the answer about “how”should the new model dofrom the micro-level. Through six key steps, which consist of datapreparation, concept lattice construction, concept lattice analysis, tax-folk mapping, tax-folk hybrid classification tree generation, outputand evaluation, the operations procedures of grafting folksonomy totaxonomy formed a mixed classification can be truly actualized usingformal concept analysis.Select “Douban” as a case analysis object from many social taggingsystems, and select the book resources which called“books read by fiftythousand to one hundred thousand people”as experimental resources, wecan collect relevant data and give an empirical study use the specificprocess of resources aggregation based on the Tax-folk hybrid model, andanalyze the experiment result to proof whether the theory we proposed inthis paper is scientific, reasonable and workable.The research on resource aggregation based on the Tax-folk hybridmodel in social tagging system has new value not only because the newopinion of “ideal resource organization tree” is proposed, and the newways to achieve social tagging system resource aggregation is found byconstructing tax-folk hybrid model, but also because the tax-folk hybridclassification tree formed from the model has very high advantages onresource organization.It has two aspects of significance at least that we do the researchon resource aggregation based on the Tax-folk hybrid model in socialtagging system, on the one hand, there is a theoretical explorationsignificance because the research improves theory architecture ofresource aggregation in social tagging system, and on the other hand,there is a practical significance because the tax-folk hybridclassification improves the resources usage efficiency of social taggingsystem, and thus bring economic and social value.更多还原