网络驱动的未识甲骨字特性及场景语义预测

Network-driven prediction of unknown oracle character’s features and scene semantics

甲骨学的研究具有重要的文化价值和传承意义,可以极大提高国家的文化自信。未识甲骨字的语义预测是甲骨学研究中最主要的问题,也是传统甲骨学研究中最棘手的问题。现有的计算机技术辅助研究方法无法预测未识甲骨字的语义。利用复杂网络对甲骨文进行了抽象和理解,并对未识甲骨字的场景语义进行预测。首先,以甲骨拓片为基础数据,通过建模构建甲骨字网络;其次,在甲骨字网络之上,分析未识甲骨字的重要性、信息丰富度、闭合性等特性,为预测未识甲骨字的场景语义提供理论依据;最后,根据网络特性和甲骨拓片的上下文语境预测未识甲骨字的场景语义。构建的未识甲骨字特性体系以及预测未识甲骨字的场景语义思路为破译其他未识甲骨字的语义奠定了基础,有助于推动甲骨文考释的进程。

The research of oracle bone inscriptions is of important cultural values and inheritance significance,which can greatly enhance the country′s cultural self-confidence.The semantic prediction of unknown oracle characters is recognized as the most important and difficult problem in studying oracle bone inscriptions,In this paper,we use complex networks to abstract and understand oracle bone inscriptions and predict the semantics of unknown oracle characters.First,we employ the oracle bone rubbings to build the oracle character network.Then,we analyze the importance,information richness and closure of unknown oracle characters based on oracle character network,these features can provide vital theories in predicting semantics of unknown oracle characters.Finally,the scene semantics of the unknown oracle characters are predicted according to the network characteristics and context of oracle bone rubbings.The features of unknown oracle characters and the process of prediction scene semantics of the unknown oracle characters provide researchers with potential to explore new approach,and promote the process of oracle bones interpretation.