Synthetic word parsing improves Chinese word segmentation

Abstract

We present a novel solution to improve the performance of Chinese word seg-mentation (CWS) using a synthetic word parser. The parser analyses the inter-nal structure of words, and attempts to convert out-of-vocabulary words (OOVs) into in-vocabulary fine-grained sub-words. We propose a pipeline CWS system that first predicts this fine-grained segmenta-tion, then chunks the output to recon-struct the original word segmentation stan-dard. We achieve competitive results on the PKU and MSR datasets, with substan-tial improvements in OOV recall.