Neural spelling correction: translating incorrect sentences to correct sentences for multimedia

Chanjun Park; Kuekyeng Kim; Yeong Wook Yang; Minho Kang; Heuiseok Lim

doi:10.1007/s11042-020-09148-2

Neural spelling correction: translating incorrect sentences to correct sentences for multimedia

Chanjun Park, Kuekyeng Kim, Yeong Wook Yang, Minho Kang, Heuiseok Lim

Research output: Contribution to journal › Article › peer-review

12 Citations (Scopus)

Abstract

The aim of a spelling correction task is to detect spelling errors and automatically correct them. In this paper we aim to perform the Korean spelling correction task from a machine translation perspective, allowing it to overcome the limitations of cost, time and data. Based on a sequence to sequence model, the model aligns its source sentence with an ‘error filled sentence’ and its target sentence aligned to the correct counter part. Thus, ‘translating’ the error sentence to a correct sentence. For this research, we have also proposed three new data generation methods allowing the creation of multiple spelling correction parallel corpora from just a single monolingual corpus. Additionally, we discovered that applying the Copy Mechanism not only resolves the problem of overcorrection but even improves it. For this paper, we evaluated our model upon these aspects: Performance comparisons to other models and evaluation on overcorrection. The results show the proposed model to even out-perform other systems currently in commercial use.

Original language	English
Pages (from-to)	34591-34608
Number of pages	18
Journal	Multimedia Tools and Applications
Volume	80
Issue number	26-27
DOIs	https://doi.org/10.1007/s11042-020-09148-2
Publication status	Published - 2021 Nov

Bibliographical note

Funding Information:
This research was supported by the MSIT(Ministry of Science and ICT), Korea, under the ITRC(Information Technology Research Center) support program(IITP-2020-2018-0-01405) supervised by the IITP(Institute for Information & Communications Technology Planning & Evaluation) and National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIP) (No.NRF-2017M3C4A7068189). I am very grateful to my friend Yejin Jang for helping me with correcting English.

Publisher Copyright:
© 2020, Springer Science+Business Media, LLC, part of Springer Nature.

Keywords

Automatic noise generation
Copy mechanism
Korean spelling correction
Neural machine translation
Overcorrection
Transformer

ASJC Scopus subject areas

Software
Media Technology
Hardware and Architecture
Computer Networks and Communications

Access to Document

10.1007/s11042-020-09148-2

Cite this

@article{edb7a514a61b410c8bf7b870e1c773bf,

title = "Neural spelling correction: translating incorrect sentences to correct sentences for multimedia",

abstract = "The aim of a spelling correction task is to detect spelling errors and automatically correct them. In this paper we aim to perform the Korean spelling correction task from a machine translation perspective, allowing it to overcome the limitations of cost, time and data. Based on a sequence to sequence model, the model aligns its source sentence with an {\textquoteleft}error filled sentence{\textquoteright} and its target sentence aligned to the correct counter part. Thus, {\textquoteleft}translating{\textquoteright} the error sentence to a correct sentence. For this research, we have also proposed three new data generation methods allowing the creation of multiple spelling correction parallel corpora from just a single monolingual corpus. Additionally, we discovered that applying the Copy Mechanism not only resolves the problem of overcorrection but even improves it. For this paper, we evaluated our model upon these aspects: Performance comparisons to other models and evaluation on overcorrection. The results show the proposed model to even out-perform other systems currently in commercial use.",

keywords = "Automatic noise generation, Copy mechanism, Korean spelling correction, Neural machine translation, Overcorrection, Transformer",

author = "Chanjun Park and Kuekyeng Kim and Yang, {Yeong Wook} and Minho Kang and Heuiseok Lim",

note = "Funding Information: This research was supported by the MSIT(Ministry of Science and ICT), Korea, under the ITRC(Information Technology Research Center) support program(IITP-2020-2018-0-01405) supervised by the IITP(Institute for Information & Communications Technology Planning & Evaluation) and National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIP) (No.NRF-2017M3C4A7068189). I am very grateful to my friend Yejin Jang for helping me with correcting English. Publisher Copyright: {\textcopyright} 2020, Springer Science+Business Media, LLC, part of Springer Nature.",

year = "2021",

month = nov,

doi = "10.1007/s11042-020-09148-2",

language = "English",

volume = "80",

pages = "34591--34608",

journal = "Multimedia Tools and Applications",

issn = "1380-7501",

publisher = "Springer",

number = "26-27",

}

TY - JOUR

T1 - Neural spelling correction

T2 - translating incorrect sentences to correct sentences for multimedia

AU - Park, Chanjun

AU - Kim, Kuekyeng

AU - Yang, Yeong Wook

AU - Kang, Minho

AU - Lim, Heuiseok

N1 - Funding Information: This research was supported by the MSIT(Ministry of Science and ICT), Korea, under the ITRC(Information Technology Research Center) support program(IITP-2020-2018-0-01405) supervised by the IITP(Institute for Information & Communications Technology Planning & Evaluation) and National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIP) (No.NRF-2017M3C4A7068189). I am very grateful to my friend Yejin Jang for helping me with correcting English. Publisher Copyright: © 2020, Springer Science+Business Media, LLC, part of Springer Nature.

PY - 2021/11

Y1 - 2021/11

N2 - The aim of a spelling correction task is to detect spelling errors and automatically correct them. In this paper we aim to perform the Korean spelling correction task from a machine translation perspective, allowing it to overcome the limitations of cost, time and data. Based on a sequence to sequence model, the model aligns its source sentence with an ‘error filled sentence’ and its target sentence aligned to the correct counter part. Thus, ‘translating’ the error sentence to a correct sentence. For this research, we have also proposed three new data generation methods allowing the creation of multiple spelling correction parallel corpora from just a single monolingual corpus. Additionally, we discovered that applying the Copy Mechanism not only resolves the problem of overcorrection but even improves it. For this paper, we evaluated our model upon these aspects: Performance comparisons to other models and evaluation on overcorrection. The results show the proposed model to even out-perform other systems currently in commercial use.

AB - The aim of a spelling correction task is to detect spelling errors and automatically correct them. In this paper we aim to perform the Korean spelling correction task from a machine translation perspective, allowing it to overcome the limitations of cost, time and data. Based on a sequence to sequence model, the model aligns its source sentence with an ‘error filled sentence’ and its target sentence aligned to the correct counter part. Thus, ‘translating’ the error sentence to a correct sentence. For this research, we have also proposed three new data generation methods allowing the creation of multiple spelling correction parallel corpora from just a single monolingual corpus. Additionally, we discovered that applying the Copy Mechanism not only resolves the problem of overcorrection but even improves it. For this paper, we evaluated our model upon these aspects: Performance comparisons to other models and evaluation on overcorrection. The results show the proposed model to even out-perform other systems currently in commercial use.

KW - Automatic noise generation

KW - Copy mechanism

KW - Korean spelling correction

KW - Neural machine translation

KW - Overcorrection

KW - Transformer

UR - http://www.scopus.com/inward/record.url?scp=85087374483&partnerID=8YFLogxK

U2 - 10.1007/s11042-020-09148-2

DO - 10.1007/s11042-020-09148-2

M3 - Article

AN - SCOPUS:85087374483

SN - 1380-7501

VL - 80

SP - 34591

EP - 34608

JO - Multimedia Tools and Applications

JF - Multimedia Tools and Applications

IS - 26-27

ER -

Neural spelling correction: translating incorrect sentences to correct sentences for multimedia

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this