随着网络技术的发展和普及，人类学习方式正从印刷时代步入信息时代，知识的内涵被重新定义，生产与传播过程趋于融合，交互成为重要的学习环节，以cMOOC为代表的联通主义学习形式应运而生并受到广泛关注。作为一种新型课程形态，其设计、评价和工具开发尚不完善，基础规律的认识成为指导优化实践的基础和前提。其中，社会网络和概念网络的关系正是联通主义尚未解决的基础规律问题，仍有待大量实证研究探索，且国内第一门cMOOC的成功实践提供了数据支撑和研究土壤，成为挖掘联通主义基础规律的契机。

基于以上背景，本研究要解决的核心问题为cMOOC中个体网络地位与其概念网络特征水平的相关关系，旨在从个体层面验证联通主义学习中社会网络与概念网络的相关关系，为今后开展cMOOC教学、评价和工具设计提供建议。包括三个子问题：（1）如何抽取、构建表征cMOOC的概念网络？（2）应从哪些特征维度及指标分析cMOOC概念网络？（3）cMOOC中个体社会网络地位与其概念网络特征水平是否相关？若相关，具有怎样的相关关系？

为了解决上述研究问题，本研究首先对“概念网络”、“cMOOC”等术语进行了概念界定，继而通过梳理国内外联通主义研究进展明确本研究的研究定位与价值，通过梳理个体层面的社会网络研究以及其他领域概念网络研究，了解个体网络地位和概念网络的研究成果，为本研究构建社会网络和概念网络、选取分析指标提供借鉴。

为了评估cMOOC中个体的网络地位和概念网络特征，为第三阶段开展基于数据的相关分析提供理论依据和方法指导，本研究第二阶段构建了三个理论模型：首先基于对国内外复杂网络研究中节点重要性评估指标及角色分析指标的梳理，结合联通主义理论观点和cMOOC学习情境特点，构建出了cMOOC个体网络地位评估指标体系，该指标体系包含个体重要性评估（6个指标）和个体网络角色分析（2个指标）两大维度；其次，基于已有的概念网络分析思路，借鉴概念图评价的相关成果，提出了cMOOC个体概念网络特征分析模型，该模型包含概念丰富度、概念质量（核心概念数量、核心概念质量）、概念连接（连接密度、连接深度）和概念解释（平均认知参与度、最高认知参与度）等4个维度和7个指标；最后为了对个体交互过程中的认知参与度进行评定，整合已有模型构建了联通主义学习中基于交互行为的认知参与度评定框架，将认知参与度水平划分为8个层级。

本研究第三阶段遵循数据密集型的研究范式，选用国内第一门cMOOC《互联网+教育：理论与实践的对话》第一期课程作为研究对象，运用复杂网络分析、主题聚类算法、内容分析、相关分析等方法，基于五个主题的10568条交互数据，依据第二阶段构建的理论模型开展了个体网络地位评估和概念网络特征分析，在此基础上一一分析了6个个体重要性指标和7个概念网络特征指标的相关性及相关趋势，以及个体网络角色与其概念网络特征水平的关系，得出以下5点结论：（1）6个个体重要性指标与7个个体概念网络特征指标均存在显著相关关系，就单一指标而言，相对出度和特征向量中心度更能反映个体概念网络特征水平；（2）从指标间的相关趋势来看，部分指标与个体概念网络特征指标的相关趋势呈现分段上升现象，在构建关系方程时需分段讨论；（3）个体网络地位能够反映其概念网络特征水平，但单一指标在反映个体概念网络特征时存在局限性，为了增加评估准确性，需综合多个个体重要性指标进行表征；（4）从概念解释维度看，cMOOC交互中存在两类特殊个体，一类是交互数量少但内容质量较高的个体，另一类是社会网络地位极高，碎片化观点和社会化交往内容多，未能产出更高质量内容的个体；（5）并非所有承担结构洞和中间人的个体都能生成高水平的概念网络，但影响力高、影响范围大的结构洞或中间人，其概念网络特征水平往往较高。基于研究结论进一步提出3点实践建议：（1）课程设计要关注交互活动、交互策略和激励机制的设计；（2）开展适应性、反馈性的学习评价，综合个体网络地位和概念网络特征的多项指标，设计多元化课程证书；（3）引入推荐机制，推荐重要性和影响力高的个体和关注内容相似的个体帮助学习者有效寻径。

本研究主要有三大创新点：提出了一种构建cMOOC概念网络的思路，为后续开展联通主义概念网络的相关研究提供了借鉴；构建了适用于联通主义学习情境、多维度的个体概念网络特征分析的模型，为从个体层面开展概念网络分析提供了理论依据和模型参考；首次基于实证数据从个体层面验证了cMOOC中社会网络和概念网络的相关关系，发现了能够表征个体概念网络各维度特征的个体重要性指标，为今后进一步揭示联通主义学习中两大网络之间的作用机制奠定了基础，同时也提供了一种用社会网络指标表征个体概念网络特征水平的学习评价思路。

未来可基于本研究的研究发现和研究成果，进一步探索和优化cMOOC概念网络的抽取和构建方法，进一步丰富和发展个体概念网络特征分析模型的维度和指标，在更多真实情境中应用分析开展多轮验证，构建运用社会网络地位指标评估其概念网络特征水平的算法或模型，从群体、集体等更多视角进一步探索联通主义两大网络之间的影响机制和作用规律。

With the development and popularization of network technology, human learning is entering the information age from the printing era, the knowledge is being redefined, the knowledge production and transmission are converging, and interaction is becoming more and more important in learning. Connectivist learning, represented by cMOOC, came into being and received extensive attention. As a new type of online learning, theoretical research of basic rules become the prerequisite for the design, practice and evaluation of cMOOC. And the relationship between the social network and concept network, which is one of the fundamental theoretical problems in Connectivism, need to be solved through a lot of empirical research. Meanwhile, the successful practice of the first domestic cMOOC, where there are much data supporting research, gives opportunity to solve this problem.

Based on the above background, the core research question in this study is the correlation relationship between individual social network status and corresponding concept network characteristics in cMOOC, in order to verify the correlation between social network and concept network in connectivist learning at the individual level, and to provide some advice of teaching, evaluation and tool design in cMOOC. There are three sub-questions: (1) How to construct the concept network in cMOOC? (2) Which dimensions and indicators can be used to analyze the characteristic level of the concept network in cMOOC? (3) Is the individual social network status related to the concept network characteristics in cMOOC? If so, what kind of correlation?

To solve these questions, this study first defined the terms such as "concept network" and "cMOOC". Then, comb the domestic and foreign research progress of Connectivism, to clarify the value of this study. Comb network research and concept network research in other fields, to find research results of individual social network status and concept network analysis, which can provide references for the construction of network and the selection of analysis indicators in this study.

In order to evaluate the individual social network status and concept network characteristic in cMOOC, which can provide theoretical basis and method guidance for the correlation analysis in the third stage, three theoretical models were constructed in the second phase of this study. Firstly, based on the indicators of individual importance evaluation in the research of complex networks, combined with the views of Connectivism and the characteristics of cMOOC, this study proposed an evaluation framework of individual social network status in cMOOC, which includes the two dimensions of individual importance assessment (6 indicators) and individual network role analysis (2 indicators); secondly, based on the existing concept network analysis ideas, and drawing on the relevance of concept map evaluation, this study proposed a model for individual concept network characteristics analysis in cMOOC, which includes 4 dimensions and 7 indicators of Concept Richness, Concept Quality (Number of Keywords, Quality of Keywords), Concept Connection (Number of Connections, Connection Depth), and Concept Explanation (Average Cognitive Engagement, Highest Cognitive Engagement); finally, in order to evaluate the individual cognitive engagement  in the process of interaction, based on the CIE model, this study also proposed an evaluation framework of cognitive engagement based on interactive behavior in connectivist learning, in which the level of cognitive engagement was divided into 8 levels.

Guided by data-intensive research paradigm, the first domestic cMOOC "Internet plus Education: Dialogue between Theory and Practice" was selected as the research object. Using complex network analysis, topic clustering algorithms, content analysis and correlation analysis method, based on three theoretical model, this study collected 10568 interactive data to analyze the individual network status and concept network characteristic analysis, and then analyzed the correlation relationship between them. This study drew the following 5 conclusions: (1) there are significant correlations between 6 individual importance indicators and 7 concept network characteristic indicators. For a single indicator, the out-degree and eigenvector centrality can better reflect the characteristic level of the individual concept network; (2) some correlation trends between indicators show a piecewise rising phenomenon, which needs to be discussed in sections when constructing the relationship equation; (3) individuals network status can reflect the concept network characteristic, but a single indicator has limitations. In order to increase the accuracy, multiple individual importance indicators must be synthesized to represent it; (4) from the perspective of Concept Explanation, there are two types of special individuals in cMOOC, some individuals have less interactions with high content quality, and some have high network status but fail to produce higher quality content; (5) Not all individuals who are structural holes and middlemen in social network can generate high-level concept network, but much higher the influence is, much higher the concept network quality is. Based on the research conclusions, three further practical suggestions were put forward: (1) cMOOC design should focus on the design of interactive activities, interactive strategies and incentive mechanisms; (2) carry out adaptive and feedback-based learning evaluation, integrate multiple indicators of individual network status and the concept network characteristics, and design diversified curriculum certificate; (3) design recommendation mechanism to recommend individuals with high importance and influence and similar concerns, to help learners make effective way-finding.

There are three major innovations of this study: firstly, this study proposed a new idea of concept network construction, which provides a reference for the subsequent research on the concept network in connectivist learning; secondly, this study formed a multi-dimensional model of individual concept network characteristics analysis for connectivist learning, which provides theoretical basis and model reference for conceptual network analysis at the individual level; finally, it’s the first time that verify the correlation between social network and concept network in connectivist learning at the individual level based on data analysis, which lays the foundation for further revealing the action mechanism  between the two networks in connectivist learning, and also provides a new idea of concept network characteristics evaluation using social network indicators in connectivist learning.

Based on this study, we can further explore and optimize the construction method of concept network in cMOOC, develop the dimensions and indicators of the individual concept network characteristic analysis model, collect much data to verify the research results in more real situations, build an algorithm or model that uses social network status indicators to assess concept network characteristics, and further explore the impact mechanism of the two networks in connectivist learning from more perspectives such as groups and collectives.