A Hybrid Model for Chinese Spelling Check

Abstract

Spelling check for Chinese has more challenging difficulties than that for other languages. A hybrid model for Chinese spelling check is presented in this article. The hybrid model consists of three components: one graph-based model for generic errors and two independently trained models for specific errors. In the graph model, a directed acyclic graph is generated for each sentence, and the single-source shortest-path algorithm is performed on the graph to detect and correct general spelling errors at the same time. Prior to that, two types of errors over functional words (characters) are first solved by conditional random fields: the confusion of &#8220;&#22312;&#8221; (<i>at</i>) (pinyin is <i>zai</i> in Chinese), &#8220;&#20877;&#8221; (<i>again</i>, <i>more</i>, <i>then</i>) (pinyin: <i>zai</i>) and &#8220;&#30340;&#8221; (<i>of</i>) (pinyin: <i>de</i>), &#8220;&#22320;&#8221; (-<i>ly</i>, adverb-forming particle) (pinyin: <i>de</i>), and &#8220;&#24471;&#8221; (<i>so that</i>, <i>have to</i>) (pinyin: <i>de</i>). Finally, a rule-based model is exploited to distinguish pronoun usage confusion: &#8220;&#22905;&#8221; (<i>she</i>) (pinyin: <i>ta</i>), &#8220;&#20182;&#8221; (<i>he</i>) (pinyin: <i>ta</i>), and some other common collocation errors. The proposed model is evaluated on the standard datasets released by the SIGHAN Bake-off shared tasks, giving state-of-the-art results.

DOI: 10.1145/3047405

20 Figures and Tables

Cite this paper

@article{Zhao2017AHM, title={A Hybrid Model for Chinese Spelling Check}, author={Hai Zhao and Deng Cai and Yang Xin and Yuzhu Wang and Zhongye Jia}, journal={ACM Trans. Asian & Low-Resource Lang. Inf. Process.}, year={2017}, volume={16}, pages={21:1-21:22} }