基于大规模真实语料汉语词汇联想意义网络的构建——纪念吕叔湘先生百年诞辰

文章提出了一个利用大规模真实语料自动提取汉语词汇联想意义的方法,利用“像Y那样A”和“A得像Y那样”作为提取框架作连续推演,将生成一个体词和谓词构成的复杂关系系统。提取的联想意义构成了一个封闭而又无限的神经网络,网络每一个节点的“意义”实际上是由该节点所连接的弧线所定义,节点词语的意义就是各弧线另一端点意象的复合,这种意义是真实约,而传统义素分析法所流行的二分树形的义素组合则是虚拟的,而且具有很大的随意性。文章构建了一个汉语词汇联想意义客观的、量化的、系统的多维模型。

The Construction of a Semantic Network of Chinese Lexical Associations Based on Large-Scale Corpora

This study proposes an approach for the automatic extraction of lexical associations in the Chinese language based on analyzing large-scale corpora. With the progressive pattern matching of the phrasal frames "xiang4 Y na4yang4 A (like Y (so/as such) A)" and "A de xiang4 Y na4yang4" (A CSC * like Y so), a complex relational system of objects and predicates, which takes the form of a closed and infinite neural network, can be established for representing lexical associations. The meaning represented by each node in the network is defined by the arcs extending from it. The meaning of the lexeme on the node is the holistic combination of all meanings represented by its immediate neighbors, i.e. all of its lexical associations. Meaning analyzed as such is based on real-world language use, while meaning combinations in a binary tree in the traditional semantic componential analysis result from a priori considerations and could be arbitrary. This study tries to construct an authentic, objective and tractable multi-dimensional model for lexical associations in the Chinese language. CSC: Complex Stative Construction