英语网络非正规语言表达的自动识别与术语库构建

Automatic Recognition and Terminology Database Construction of English Network Informal Language Expressions

网络非正规语言表达(Network Informal Language Expression, NILE)具有的创新性强、超常规化、表达方式口语化等特点,为许多自然语言处理任务带来了挑战。在使用网络语言进行交流的过程中,部分网络非正规语言表达逐渐标准化和规范化,形成网络非正规语言表达术语。通过对46万余条Twitter数据的收集、处理和分析,英语网络非正规语言表达可以在音、形、义层面划分为13类,并对其特征进行分析和总结。结合统计方法和规则方法的优点,设计了统计和规则融合的英语网络非正规语言表达自动识别方法。最终构建规模为7000余条的网络非正规语言表达术语库。

Network Informal Language Expression(NILE) has the characteristics of novelty, unconventionality and colloquialism, which poses a challenge to many natural language processing tasks. In the process of using online language for communication, some NILEs are gradually standardized and normalized, forming a crucial part of the NILE terminology. By collecting, processing and analyzing more than 460 000 tweets, we divide English NILEs into 13 categories from the perspectives of sound, form and sense, and further analyzed their characteristics. Taking the advantage of statistic-based approach and rule-based approach, we design an automatic English NILE recognition system based on the integration of statistical techniques and linguistic rules, and thereupon build a terminology database of 7000 NILE items.