Resumen de New Light Shed on Chinese Word Segmentation in MT by a Language Investigation

Zhijie Wu

• The Chinese language, unlike some western languages, is written without a space between any two words, which presents itself as a unique problem in Machine Translation: how to segment words in Chinese? The current word-segmentation systems in Machine Translation are either linguistically-oriented or statistically-oriented. Both types, however, have some innate defects that cannot be overcome due to the pragmatically-oriented feature of the Chinese language. This research aims at addressing the problem of Chinese word segmentation of Machine Translation in light of a language investigation consisting of two surveys and eight interviews.

  
  
  Plan de l'article
  
  1. Introduction
  2. Surveys, Interviews, Experiments and their Findings
  2.1 Surveys
  2.1.1 The First Survey
  2.1.1.1 Subjects
  2.1.1.2 Questionnaires
  2.1.1.3 Data Collection
  2.1.1.4 Data Analysis
  2.1.2 The Follow-up Survey
  2.1.2.1 Subjects
  2.1.2.2 Questionnaires
  2.1.2.3 Data Collection
  2.1.2.4 Data Analysis
  2.2 Interviews
  2.2.1 Subjects
  2.2.2 Measures
  2.2.3 Data Collection
  2.2.4 Data Analysis
  2.3 Experiments
  2.3.1 Materials
  2.3.2 Instruments
  2.3.3 Data Collection
  2.3.4 Data Analysis
  3. Findings and Their Implications
  3.1 Potentiality for Machines to Carry Out Word Segmentation
  3.2 Contextual Information in Word Segmentation
  3.3 Semantic Plausibility as the Criterion of Word Segmentation
  3.4 Word Segmentation: a Means Rather Than a Purpose
  3.1.5 Implications