Utilizing crowdsourcing for the construction of chinese-mongolian speech corpus with evaluation mechanism

Rihai Su; Shumin Shi; Meng Zhao; Heyan Huang

doi:10.1007/978-981-10-6388-6_5

Utilizing crowdsourcing for the construction of chinese-mongolian speech corpus with evaluation mechanism

Rihai Su, Shumin Shi^*, Meng Zhao, Heyan Huang

^*此作品的通讯作者

计算机学院

科研成果: 书/报告/会议事项章节 › 会议稿件 › 同行评审

2 引用（Scopus）

摘要

Crowdsourcing has been used recently as an alternative to traditional costly annotation by many natural language processing groups. In this paper, we explore the use of Wechat Official Account Platform (WOAP) in order to build a speech corpus and to assess the feasibility of using WOAP followers (also known as contributors) to assemble speech corpus of Mongolian. A Mongolian language qualification test was used to filter out potential non-qualified participants. We gathered natural speech recordings in our daily life, and constructed a Chinese-Mongolian Speech Corpus (CMSC) of 31472 utterances from 296 native speakers who are fluent in Mongolian, totalling 30.8 h of speech. Then, an evaluation experiment was performed, in where the contributors were asked to choose a correct sentence from a multiple choice list to ensure the high-quality of corpus. The results obtained so far showed that crowdsourcing for constructing CMSC with an evaluation mechanism could be more effective than traditional experiments requiring expertise.

源语言	英语
主期刊名	Data Science - 3rd International Conference of Pioneering Computer Scientists, Engineers and Educators, ICPCSEE 2017, Proceedings
编辑	Qilong Han, Beiji Zou, Xiaoning Peng, Zeguang Lu, Guanglu Sun, Weipeng Jing
出版商	Springer Verlag
页	55-65
页数	11
ISBN（印刷版）	9789811063879
DOI	https://doi.org/10.1007/978-981-10-6388-6_5
出版状态	已出版 - 2017
活动	3rd International Conference of Pioneer Computer Scientists, Engineers, and Educators, ICPCSEE 2017 - Changsha, 中国期限: 22 9月 2017 → 24 9月 2017

出版系列

姓名	Communications in Computer and Information Science
卷	728
ISSN（印刷版）	1865-0929

会议

会议	3rd International Conference of Pioneer Computer Scientists, Engineers, and Educators, ICPCSEE 2017
国家/地区	中国
市	Changsha
时期	22/09/17 → 24/09/17

访问文件

10.1007/978-981-10-6388-6_5

其它文件与链接

链接到 Scopus 的出版物

引用此

Su, R., Shi, S., Zhao, M., & Huang, H. (2017). Utilizing crowdsourcing for the construction of chinese-mongolian speech corpus with evaluation mechanism. 在 Q. Han, B. Zou, X. Peng, Z. Lu, G. Sun, & W. Jing (编辑), Data Science - 3rd International Conference of Pioneering Computer Scientists, Engineers and Educators, ICPCSEE 2017, Proceedings (页码 55-65). (Communications in Computer and Information Science; 卷 728). Springer Verlag. https://doi.org/10.1007/978-981-10-6388-6_5

Su, Rihai ; Shi, Shumin ; Zhao, Meng 等. / Utilizing crowdsourcing for the construction of chinese-mongolian speech corpus with evaluation mechanism. Data Science - 3rd International Conference of Pioneering Computer Scientists, Engineers and Educators, ICPCSEE 2017, Proceedings. 编辑 / Qilong Han ; Beiji Zou ; Xiaoning Peng ; Zeguang Lu ; Guanglu Sun ; Weipeng Jing. Springer Verlag, 2017. 页码 55-65 (Communications in Computer and Information Science).

@inproceedings{5c29432c32ea4a7687f48dd36808bc2d,

title = "Utilizing crowdsourcing for the construction of chinese-mongolian speech corpus with evaluation mechanism",

abstract = "Crowdsourcing has been used recently as an alternative to traditional costly annotation by many natural language processing groups. In this paper, we explore the use of Wechat Official Account Platform (WOAP) in order to build a speech corpus and to assess the feasibility of using WOAP followers (also known as contributors) to assemble speech corpus of Mongolian. A Mongolian language qualification test was used to filter out potential non-qualified participants. We gathered natural speech recordings in our daily life, and constructed a Chinese-Mongolian Speech Corpus (CMSC) of 31472 utterances from 296 native speakers who are fluent in Mongolian, totalling 30.8 h of speech. Then, an evaluation experiment was performed, in where the contributors were asked to choose a correct sentence from a multiple choice list to ensure the high-quality of corpus. The results obtained so far showed that crowdsourcing for constructing CMSC with an evaluation mechanism could be more effective than traditional experiments requiring expertise.",

keywords = "Chinese-Mongolian corpus, Crowdsourcing, Evaluation mechanism, Speech corpus, WOAP",

author = "Rihai Su and Shumin Shi and Meng Zhao and Heyan Huang",

note = "Publisher Copyright: {\textcopyright} 2017, Springer Nature Singapore Pte Ltd.; 3rd International Conference of Pioneer Computer Scientists, Engineers, and Educators, ICPCSEE 2017 ; Conference date: 22-09-2017 Through 24-09-2017",

year = "2017",

doi = "10.1007/978-981-10-6388-6_5",

language = "English",

isbn = "9789811063879",

series = "Communications in Computer and Information Science",

publisher = "Springer Verlag",

pages = "55--65",

editor = "Qilong Han and Beiji Zou and Xiaoning Peng and Zeguang Lu and Guanglu Sun and Weipeng Jing",

booktitle = "Data Science - 3rd International Conference of Pioneering Computer Scientists, Engineers and Educators, ICPCSEE 2017, Proceedings",

address = "Germany",

}

Su, R, Shi, S, Zhao, M & Huang, H 2017, Utilizing crowdsourcing for the construction of chinese-mongolian speech corpus with evaluation mechanism. 在 Q Han, B Zou, X Peng, Z Lu, G Sun & W Jing (编辑), Data Science - 3rd International Conference of Pioneering Computer Scientists, Engineers and Educators, ICPCSEE 2017, Proceedings. Communications in Computer and Information Science, 卷 728, Springer Verlag, 页码 55-65, 3rd International Conference of Pioneer Computer Scientists, Engineers, and Educators, ICPCSEE 2017, Changsha, 中国, 22/09/17. https://doi.org/10.1007/978-981-10-6388-6_5

Utilizing crowdsourcing for the construction of chinese-mongolian speech corpus with evaluation mechanism. / Su, Rihai; Shi, Shumin; Zhao, Meng 等.
Data Science - 3rd International Conference of Pioneering Computer Scientists, Engineers and Educators, ICPCSEE 2017, Proceedings. 编辑 / Qilong Han; Beiji Zou; Xiaoning Peng; Zeguang Lu; Guanglu Sun; Weipeng Jing. Springer Verlag, 2017. 页码 55-65 (Communications in Computer and Information Science; 卷 728).

科研成果: 书/报告/会议事项章节 › 会议稿件 › 同行评审

TY - GEN

T1 - Utilizing crowdsourcing for the construction of chinese-mongolian speech corpus with evaluation mechanism

AU - Su, Rihai

AU - Shi, Shumin

AU - Zhao, Meng

AU - Huang, Heyan

PY - 2017

Y1 - 2017

N2 - Crowdsourcing has been used recently as an alternative to traditional costly annotation by many natural language processing groups. In this paper, we explore the use of Wechat Official Account Platform (WOAP) in order to build a speech corpus and to assess the feasibility of using WOAP followers (also known as contributors) to assemble speech corpus of Mongolian. A Mongolian language qualification test was used to filter out potential non-qualified participants. We gathered natural speech recordings in our daily life, and constructed a Chinese-Mongolian Speech Corpus (CMSC) of 31472 utterances from 296 native speakers who are fluent in Mongolian, totalling 30.8 h of speech. Then, an evaluation experiment was performed, in where the contributors were asked to choose a correct sentence from a multiple choice list to ensure the high-quality of corpus. The results obtained so far showed that crowdsourcing for constructing CMSC with an evaluation mechanism could be more effective than traditional experiments requiring expertise.

AB - Crowdsourcing has been used recently as an alternative to traditional costly annotation by many natural language processing groups. In this paper, we explore the use of Wechat Official Account Platform (WOAP) in order to build a speech corpus and to assess the feasibility of using WOAP followers (also known as contributors) to assemble speech corpus of Mongolian. A Mongolian language qualification test was used to filter out potential non-qualified participants. We gathered natural speech recordings in our daily life, and constructed a Chinese-Mongolian Speech Corpus (CMSC) of 31472 utterances from 296 native speakers who are fluent in Mongolian, totalling 30.8 h of speech. Then, an evaluation experiment was performed, in where the contributors were asked to choose a correct sentence from a multiple choice list to ensure the high-quality of corpus. The results obtained so far showed that crowdsourcing for constructing CMSC with an evaluation mechanism could be more effective than traditional experiments requiring expertise.

KW - Chinese-Mongolian corpus

KW - Crowdsourcing

KW - Evaluation mechanism

KW - Speech corpus

KW - WOAP

UR - http://www.scopus.com/inward/record.url?scp=85030098362&partnerID=8YFLogxK

U2 - 10.1007/978-981-10-6388-6_5

DO - 10.1007/978-981-10-6388-6_5

M3 - Conference contribution

AN - SCOPUS:85030098362

SN - 9789811063879

T3 - Communications in Computer and Information Science

SP - 55

EP - 65

BT - Data Science - 3rd International Conference of Pioneering Computer Scientists, Engineers and Educators, ICPCSEE 2017, Proceedings

A2 - Han, Qilong

A2 - Zou, Beiji

A2 - Peng, Xiaoning

A2 - Lu, Zeguang

A2 - Sun, Guanglu

A2 - Jing, Weipeng

PB - Springer Verlag

T2 - 3rd International Conference of Pioneer Computer Scientists, Engineers, and Educators, ICPCSEE 2017

Y2 - 22 September 2017 through 24 September 2017

ER -

Su R, Shi S, Zhao M, Huang H. Utilizing crowdsourcing for the construction of chinese-mongolian speech corpus with evaluation mechanism. 在 Han Q, Zou B, Peng X, Lu Z, Sun G, Jing W, 编辑, Data Science - 3rd International Conference of Pioneering Computer Scientists, Engineers and Educators, ICPCSEE 2017, Proceedings. Springer Verlag. 2017. 页码 55-65. (Communications in Computer and Information Science). doi: 10.1007/978-981-10-6388-6_5

Utilizing crowdsourcing for the construction of chinese-mongolian speech corpus with evaluation mechanism

摘要

出版系列

会议

访问文件

其它文件与链接

指纹

引用此