A head-driven approach for automatic identification of Chinese coordinate structures

Automatic identification of coordinate structure is a challenging task for sentence analysis in natural language processing. This paper presents a head-driven approach to automatically identify Chinese coordinate structures. The basic idea can be outlined as "boundary markers + head similarity + structure parallelism". Test on 2,400 coordinate structures shows the precision up to 78.0%, which is strikingly better than the string-similarity-based method reported in previous works.Computer Science, Artificial IntelligenceCPCI-S(ISTP)