论文部分内容阅读
汉语复句层次关系的自动识别是中文信息处理的难点之一,而其中关系标记非充盈模式下无标分句的层次归属以及关系标记充盈模式下某些特殊因果类、并列类有标复句的分句关联的确定更是难中之难。文章提出利用分句语义关联理论来解决这类问题:1)从句法和语义两个角度,总结出标示分句语义关联的3大类14个小类;2)根据特征之间的相互约束规律和语料中分句特征的统计规律,确定特征分析的先后顺序;3)讨论分句间语义关联度的计算方法,根据分句间的关联度,确定分句的层次归属。
The automatic recognition of Chinese complex sentence hierarchy is one of the difficulties in Chinese information processing. However, the hierarchical attribution of unmarked clauses in the non-full-fledged pattern and some special causal patterns in the relational fills pattern, Sentence association is more difficult to determine. The article proposes to use the clause semantic association theory to solve such problems: 1) From the perspectives of syntax and semantics, the author concludes 14 sub-categories of 3 major categories which indicate the semantic association of sentences; 2) According to the law of mutual restraint And the corpus of statistical characteristics of sentence features to determine the order of the characteristics of the analysis; 3) to discuss the calculation of the semantic relevance between clauses, according to the relevance of clauses to determine the level of clauses attribution.