55_用语图分析揭示语言系统中的隐性规律.pdf

Confirmation Number: 55 Submission Passcode: 55X-A3H9F8D4P6 用语图分析揭示语言系统中的隐性规律 ——赢家通吃和赢多输少算法 1 2 陈振宁 ，陈振宇 （1.浙江大学，浙江杭州，310058；2.复旦大学，上海，200433） 摘 要：本文用“图”这一数学工具，通过定量分析来揭示语言系统中的隐性规律。设计了赢家通吃和赢 多输少两种生成算法，将理想算法 “步步竞争、择优而行”的博弈论思路贯彻到非理想状态。两种新算法 都较前人有更好的概括能力。赢多输少算法更兼顾了充分概括和适度概括均衡。生成语图后设计着重准确 率的最小简图和着重覆盖率的最大简图归纳算法，挖掘控制的主流规则、分析语言系统的语言学规律。在 最小简图基础上提出控制度公式以评价语言系统。 关键词：隐性规律，图论，博弈论，规则挖掘 中图分类号：H030 文献标识码：A Revealing Covert Laws in Language Systems through Graphs ——Winner-Get-All Winner-More-Loser-Less Chen Zhenning, Chen Zhenyu (1. Zhejiang University, Hangzhou, Zhejiang, 310058, China; 2. Fudan University, Shanghai, 200433, China) Abstract: We tried to reveal convert laws with quantitative analysis through graphs and designed two generating algorithms of language graphs: Winner-get-all and Winner-more-loser-less, which extrapolated game theory used by idea-algorithm to none-perfect state. Given consideration to full generation, our algorithms are more powerful than algorithms made before. Furthermore, we created a balance between full and modest generation in the Winner-more-loser-less algorithm. There are two kinds of inductive algorithms to mine mainstream rules and analyze linguistic laws: Min-Subgraphs for accuracy, as well as Max-Subgraphs for coverage. A formula for control degree based on min-subgraphs was put forward to evaluate language systems. Keywords: Covert Laws; Graph Theory; Game Theory; Rules Mining 1 引论： 类型学在跨语言比较研究中引入语义地图（Semantic Maps）理论，它以基元（即所调 查的语言项目）为“点（node ）”，根据这些项基元的共现“关系（relationship ）”连接成“边 (edge) ”，生成一个“图（graph ）”。然后用这一地图去挖掘各项目间的规律。[1][2][3]这种地图