Fast and Accurate Neural Word Segmentation for Chinese

TL;DR

This paper introduces a fast, accurate, end-to-end neural Chinese word segmenter that improves efficiency and performance over existing models by balancing word and character embeddings.

Contribution

The paper proposes a novel greedy neural segmentation model with balanced embeddings, enhancing speed and accuracy for Chinese word segmentation.

Findings

Outperforms state-of-the-art neural models in speed and accuracy

Achieves competitive results on Chinese benchmark datasets

Reduces computational inefficiency of previous neural models

Abstract

Neural models with minimal feature engineering have achieved competitive performance against traditional methods for the task of Chinese word segmentation. However, both training and working procedures of the current neural models are computationally inefficient. This paper presents a greedy neural word segmenter with balanced word and character embedding inputs to alleviate the existing drawbacks. Our segmenter is truly end-to-end, capable of performing segmentation much faster and even more accurate than state-of-the-art neural models on Chinese benchmark datasets.