期刊
NEURAL COMPUTING & APPLICATIONS
卷 34, 期 23, 页码 20629-20638出版社
SPRINGER LONDON LTD
DOI: 10.1007/s00521-022-07489-7
关键词
Natural language processing; Tamil; LSTM; GRU; Deep learning; Spell checking
Grammar checking is an important application of Natural Language Processing, but there is a lack of grammar checkers for Tamil language. This paper proposes a hybrid approach that combines neural network and rule-based methods to develop a Tamil grammar checker, addressing issues such as spell checking, consonant errors, long component letter errors, and subject-verb agreement errors.
Grammar checking is one of the important applications of Natural Language Processing. Though the work in this area has been started decades before, the requirement of full-fledged grammar checking is still a demanding task. The recent revolution of Internet requires the computers not only deal with English Language but also in regional languages. People, who do not know English, tend to interact with computers through their regional language. Tamil is one such regional language which is recognized as classical (Semmozhi) language. Grammar checker application has been implemented for languages like English, Urdu, Punjabi, etc. But as far as Tamil is concerned, grammar checker is very scarce. There are many approaches to develop a grammar checker application. It can be statistical based, rule based or deep learning based. The proposed method involves hybrid approach to develop a Tamil grammar checker as Tamil has lot of grammatical features. In the proposed work, we concentrated on spell checking, consonant (Punarchi) error handling, long component letter error and subject-verb agreement errors. To tackle all these errors, combination of neural network approach as well as rule-based approach is proposed in this paper.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据