双向型高搭配力度ngram链在语料库建设中的应用

对搭配的研究是语言研究的焦点之一,搭配研究不仅包括两个词之间的搭配,还包括各种扩展的搭配,最常见的对搭配的扩展是搭配在频数上的扩展,通常称为词串(lexical bun-dles)。本文研究的是搭配的另外一种维度的扩展——按照搭配力度的双向扩展,称为双向型高搭配力度ngram链,即把搭配力度最强的多词序列提取出来,然后进行分析。分析发现:双向型高搭配力度ngram链中大部分序列与文本内容相关度比较高,并且结构相对完整,可以作为候选的关键词组。

Utilization of Bi-directional Ngram Chains of High Collocatabiliy in Corpus Building

The study on collocation is one of the focuses of language studies,which is not limited to collocation between two words;it encompasses extended studies of collocation,such as extension of collocation based on raw frequency,which is often called lexical bundles.In this thesis,collocation is extended on both directions(to the left and to the right) based on another criteria collocation.After analyzing the structures resulted,it is concluded that bi-directional ngram chains of high collocatabiliy are with relatively well-formed and highly content sensitive,which are qualified to be potential candidate key phrases.