从词汇语意学到知识体系： 语料中呈现的认知系统 [课件].ppt

From Linguistic Insights to Theoretical Relevance語料觀察與理論詮釋 Chu-Ren Huang Academia Sinica .tw/huang/huang.htm Or: How to avoid getting deleted? Old linguists don’t die, they just get obligatorily deleted. -Title of Eva Hajicova’s talk given at her speech when accepting the Association of Computational Linguistics Life Achievement Award in July 2006 -From an old Indiana University Linguistics Club T-Shirt How to avoid getting deleted I I have data, lots of data! And some linguistic generalizations too. How Much Data? Comparing Chinese Corpora How to find generalizations from Corpus I Size Matters Observable facts are directly proportional to corpus size An average collocation requires corpus size of 1,000,000,000 (10,000x10,000x10) Hence the Gigaword Corpus How Much Data Can One have? Chinese Gigaword Corpus Size of Chinese Gigaword Corpus Size of Lexicon Extracted from Chinese Gigaword Corpus How to find generalizations from Corpus II Numbers (and tendencies) don’t lie A large enough sample allows us to observe the tendencies (distributional patterns) of language use Linguistic Knowledge can be used to generate more linguistic knowledge In order to extract grammatical relations automatically and efficiently, we applied detail lexical grammatical knowledge, especially argument structures and collocating patterns of verbs Finding a Word’s Company: Corpus KeyWord In Context (KWIC) and the color pen approach A Word’s Company Automatically Detected: WordSketch w BNC Data Sketch Engine and Chinese WordSketch Sketch Engine http://www.sketchengine.co.uk Developed by team led by Adam Kilgarriff A new corpus viewing tool Discovering grammatical information from a gigantic corpus Chinese Wordsketch by Academia Sinica .tw/wordsketch (free license for use in Taiwan only) Academia Sinica, Taiwan (Huang, Smith, Ma, Simon黃居仁，史尚明，馬偉雲，石穆) WordSketch’s Approach:From Lexical Types to Relations Types BNC has 100,000,000 Words 939,028 word types 70,000,000 tuples (relations)