客观知识体系中基于语义域的语义关系组合分类模型研究

Research on Classification Model for Combination of Semantic Relations Based on Semantic Domain in in Objective Knowledge System

当前对客观知识体系中语义关系组合方面的研究相对较少，且目前都无法为任意两个相邻语义关系的组合提供判断依据。由于现有的客观知识体系中偶然存在着大量的有组合结果的语义关系组合样本数据的存在，使本文采用分类模型来研究语义关系的组合问题成为可能。此外，本文将语义域引入语义关系组合分类模型，为语义关系组合模型研究提供了一个新的途径。所以，本文从特征项抽取构造和模型构造两方面人手来建立基于语义域的语义关系组合模型，并从整体准确率、各个组合结果的准确率和前人结论的准确率三个方面对实验结果进行分析，归纳出了基于语义域的语义关系组合规律和研究的局限性。

Little research is known about the combination of semantic relations in the objective knowledge system before. Moreover, at present, there is no judgment basis for the combination of any two adjacent semantic relations. While there is a lot of sample data of the combination of semantic relations which has combination result in the objective knowledge system, it made it possible to study the combination of semantic relations using classification model. In addition, this paper introduced the semantic domain which provides a new way for the research on the combination of semantic relations. So this paper established the combination model of semantic relations based on the semantic domain from the two aspects ： feature extraction and the model construction, then the experimental results were analyzed from three aspects： whole rate of accuracy, the accuracy of each combination the results and the accuracy of former conclusion. In the end, combination rules and the limitations of the study based on semantic domain were generalized.