一种基于链接和语义关联的知识图示化方法

发布时间：2018-12-24 13:57

【摘要】：将海量的知识梳理成人类更容易接受的形式,一直是数据分析领域的难题.大多数传统分析方式直接对知识本身进行总结和描述概念化(conceptualization);而一些教育实践证明,从临近的知识单元进行刻画图示化(schematization)更容易使一个知识点被人类接受.在目前的经典计算机知识表达方法中,知识图示化主要依靠人工整理完成.提出了一种利用计算机自动化完成知识图示化的方法,依托维基百科概念拓扑图,探究概念与其临近概念的关系,并且提出了基于链接的自动筛选最关联概念算法;使用目前最新的神经网络模型Word2Vec对概念间的语义相似度进行量化,进一步改进关联概念算法,提高知识图示化效果.实验结果表明:基于链接的关联概念算法取得了良好的准确率,Word2Vec模型可以有效提高关联概念的排序效果.提出的方法能够准确有效地主动分析知识结构,梳理知识脉络,为科研工作者和学习者提供切实有效的建议.
[Abstract]:Combing vast amounts of knowledge into a more acceptable form has been a difficult problem in the field of data analysis. Most traditional analytical methods summarize and describe the knowledge itself directly and conceptualize (conceptualization);. Some educational practices prove that it is easier to make a knowledge point accepted by human beings by graphing (schematization) from adjacent knowledge units. In the present classical computer knowledge representation method, knowledge representation mainly depends on manual finishing. In this paper, a method of using computer automation to realize knowledge schematization is proposed, which relies on Wikipedia concept topology to explore the relationship between concept and its adjacent concept, and an algorithm for automatically selecting the most correlated concept based on link is proposed. The semantic similarity between concepts is quantified by using the latest neural network model Word2Vec to further improve the association concept algorithm and improve the effect of knowledge representation. Experimental results show that the link based association concept algorithm has good accuracy and Word2Vec model can effectively improve the ranking effect of association concepts. The proposed method can accurately and effectively analyze the knowledge structure, comb the knowledge context, and provide practical and effective advice for researchers and learners.
【作者单位】：中国科学院大学;中国科学院软件研究所;
【基金】：中国科学院系统优化基金项目(Y42901VED2,Y42901VEB1,Y42901VEB2)~~
【分类号】：TP18;TP391.1

【相似文献】