Automatic Rule-Based Correction of Stochastic Syntactic Annotation of Czech

Abstrakt

The paper presents a linguistic approach to improving the results of Czech dependency parsing. Our system based on linguistic rules identifies and corrects errors of the dependency parser (errors in dependency structure, syntactic functions or morphological tagging).

Emphasis was placed on the linguistic aspects of our correction module.

Klíčová slova

• Automatic
• Rule-Based
• Correction
• Syntactic
• Annotation
• Czech
• Dependency
• Parsing