Using knowledge graph to handle label imperfection

Y. Liu; Huakang Li; Yizheng Chen

doi:10.1007/978-3-319-13186-3_32

Using knowledge graph to handle label imperfection

Y. Liu^*, Huakang Li, Yizheng Chen

^*Corresponding author for this work

Research output: Chapter in Book or Report/Conference proceeding › Conference Proceeding › peer-review

1 Citation (Scopus)

Abstract

The performance of classification tasks extremely relies on data quality, while in real world label noises inevitably exists because of data entry errors, transmit errors and subjectivity of taggers. Different methods have been proposed to deal with label imperfection, including robust algorithms by avoid overfitting, filtering mechanism to remove noises and correction mechanism to revise noises. In this paper, we propose an approach based on knowledge graph to perceive and correct the label errors in training data. Experiments on a medical Q&A data set reveal that our knowledge graph based approach can be effective on promoting classification performance and data quality. The results as well show our approach can work in a relatively high noise level and be applied in other data mining tasks demanding deep understanding.

Original language	English
Title of host publication	Trends and Applications in Knowledge Discovery and Data Mining - PAKDD 2014 International Workshops
Subtitle of host publication	DANTH, BDM, MobiSocial, BigEC, CloudSD, MSMV-MBI, SDA, DMDA-Health, ALSIP, SocNet, DMBIH, BigPMA, Revised Selected Papers
Editors	Wen-Chih Peng, Haixun Wang, Zhi-Hua Zhou, Tu Bao Ho, Vincent S. Tseng, Arbee L.P. Chen, James Bailey
Publisher	Springer Verlag
Pages	345-356
Number of pages	12
ISBN (Electronic)	9783319131856
DOIs	https://doi.org/10.1007/978-3-319-13186-3_32
Publication status	Published - 2014
Externally published	Yes
Event	International Workshops on Data Mining and Decision Analytics for Public Health, Biologically Inspired Data Mining Techniques, Mobile Data Management, Mining, and Computing on Social Networks, Big Data Science and Engineering on E-Commerce, Cloud Service Discovery, MSMV-MBI, Scalable Dats Analytics, Data Mining and Decision Analytics for Public Health and Wellness, Algorithms for Large-Scale Information Processing in Knowledge Discovery, Data Mining in Social Networks, Data Mining in Biomedical informatics and Healthcare, Pattern Mining and Application of Big Data in conjunction with 18th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2014 - Tainan, Taiwan, Province of China Duration: 13 May 2014 → 16 May 2014

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	8643
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	International Workshops on Data Mining and Decision Analytics for Public Health, Biologically Inspired Data Mining Techniques, Mobile Data Management, Mining, and Computing on Social Networks, Big Data Science and Engineering on E-Commerce, Cloud Service Discovery, MSMV-MBI, Scalable Dats Analytics, Data Mining and Decision Analytics for Public Health and Wellness, Algorithms for Large-Scale Information Processing in Knowledge Discovery, Data Mining in Social Networks, Data Mining in Biomedical informatics and Healthcare, Pattern Mining and Application of Big Data in conjunction with 18th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2014
Country/Territory	Taiwan, Province of China
City	Tainan
Period	13/05/14 → 16/05/14

Keywords

Classification
Data quality
Knowledge graph
Label error

Access to Document

10.1007/978-3-319-13186-3_32

Cite this

Liu, Y., Li, H., & Chen, Y. (2014). Using knowledge graph to handle label imperfection. In W.-C. Peng, H. Wang, Z.-H. Zhou, T. B. Ho, V. S. Tseng, A. L. P. Chen, & J. Bailey (Eds.), Trends and Applications in Knowledge Discovery and Data Mining - PAKDD 2014 International Workshops: DANTH, BDM, MobiSocial, BigEC, CloudSD, MSMV-MBI, SDA, DMDA-Health, ALSIP, SocNet, DMBIH, BigPMA, Revised Selected Papers (pp. 345-356). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8643). Springer Verlag. https://doi.org/10.1007/978-3-319-13186-3_32

Liu, Y. ; Li, Huakang ; Chen, Yizheng. / Using knowledge graph to handle label imperfection. Trends and Applications in Knowledge Discovery and Data Mining - PAKDD 2014 International Workshops: DANTH, BDM, MobiSocial, BigEC, CloudSD, MSMV-MBI, SDA, DMDA-Health, ALSIP, SocNet, DMBIH, BigPMA, Revised Selected Papers. editor / Wen-Chih Peng ; Haixun Wang ; Zhi-Hua Zhou ; Tu Bao Ho ; Vincent S. Tseng ; Arbee L.P. Chen ; James Bailey. Springer Verlag, 2014. pp. 345-356 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{198a8c1240de46c481eb86ddcf530c6e,

title = "Using knowledge graph to handle label imperfection",

abstract = "The performance of classification tasks extremely relies on data quality, while in real world label noises inevitably exists because of data entry errors, transmit errors and subjectivity of taggers. Different methods have been proposed to deal with label imperfection, including robust algorithms by avoid overfitting, filtering mechanism to remove noises and correction mechanism to revise noises. In this paper, we propose an approach based on knowledge graph to perceive and correct the label errors in training data. Experiments on a medical Q&A data set reveal that our knowledge graph based approach can be effective on promoting classification performance and data quality. The results as well show our approach can work in a relatively high noise level and be applied in other data mining tasks demanding deep understanding.",

keywords = "Classification, Data quality, Knowledge graph, Label error",

author = "Y. Liu and Huakang Li and Yizheng Chen",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing Switzerland 2014.; International Workshops on Data Mining and Decision Analytics for Public Health, Biologically Inspired Data Mining Techniques, Mobile Data Management, Mining, and Computing on Social Networks, Big Data Science and Engineering on E-Commerce, Cloud Service Discovery, MSMV-MBI, Scalable Dats Analytics, Data Mining and Decision Analytics for Public Health and Wellness, Algorithms for Large-Scale Information Processing in Knowledge Discovery, Data Mining in Social Networks, Data Mining in Biomedical informatics and Healthcare, Pattern Mining and Application of Big Data in conjunction with 18th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2014 ; Conference date: 13-05-2014 Through 16-05-2014",

year = "2014",

doi = "10.1007/978-3-319-13186-3_32",

language = "English",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "345--356",

editor = "Wen-Chih Peng and Haixun Wang and Zhi-Hua Zhou and Ho, {Tu Bao} and Tseng, {Vincent S.} and Chen, {Arbee L.P.} and James Bailey",

booktitle = "Trends and Applications in Knowledge Discovery and Data Mining - PAKDD 2014 International Workshops",

}

Liu, Y, Li, H & Chen, Y 2014, Using knowledge graph to handle label imperfection. in W-C Peng, H Wang, Z-H Zhou, TB Ho, VS Tseng, ALP Chen & J Bailey (eds), Trends and Applications in Knowledge Discovery and Data Mining - PAKDD 2014 International Workshops: DANTH, BDM, MobiSocial, BigEC, CloudSD, MSMV-MBI, SDA, DMDA-Health, ALSIP, SocNet, DMBIH, BigPMA, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 8643, Springer Verlag, pp. 345-356, International Workshops on Data Mining and Decision Analytics for Public Health, Biologically Inspired Data Mining Techniques, Mobile Data Management, Mining, and Computing on Social Networks, Big Data Science and Engineering on E-Commerce, Cloud Service Discovery, MSMV-MBI, Scalable Dats Analytics, Data Mining and Decision Analytics for Public Health and Wellness, Algorithms for Large-Scale Information Processing in Knowledge Discovery, Data Mining in Social Networks, Data Mining in Biomedical informatics and Healthcare, Pattern Mining and Application of Big Data in conjunction with 18th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2014, Tainan, Taiwan, Province of China, 13/05/14. https://doi.org/10.1007/978-3-319-13186-3_32

Using knowledge graph to handle label imperfection. / Liu, Y.; Li, Huakang; Chen, Yizheng.
Trends and Applications in Knowledge Discovery and Data Mining - PAKDD 2014 International Workshops: DANTH, BDM, MobiSocial, BigEC, CloudSD, MSMV-MBI, SDA, DMDA-Health, ALSIP, SocNet, DMBIH, BigPMA, Revised Selected Papers. ed. / Wen-Chih Peng; Haixun Wang; Zhi-Hua Zhou; Tu Bao Ho; Vincent S. Tseng; Arbee L.P. Chen; James Bailey. Springer Verlag, 2014. p. 345-356 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8643).

Research output: Chapter in Book or Report/Conference proceeding › Conference Proceeding › peer-review

TY - GEN

T1 - Using knowledge graph to handle label imperfection

AU - Liu, Y.

AU - Li, Huakang

AU - Chen, Yizheng

N1 - Publisher Copyright: © Springer International Publishing Switzerland 2014.

PY - 2014

Y1 - 2014

N2 - The performance of classification tasks extremely relies on data quality, while in real world label noises inevitably exists because of data entry errors, transmit errors and subjectivity of taggers. Different methods have been proposed to deal with label imperfection, including robust algorithms by avoid overfitting, filtering mechanism to remove noises and correction mechanism to revise noises. In this paper, we propose an approach based on knowledge graph to perceive and correct the label errors in training data. Experiments on a medical Q&A data set reveal that our knowledge graph based approach can be effective on promoting classification performance and data quality. The results as well show our approach can work in a relatively high noise level and be applied in other data mining tasks demanding deep understanding.

AB - The performance of classification tasks extremely relies on data quality, while in real world label noises inevitably exists because of data entry errors, transmit errors and subjectivity of taggers. Different methods have been proposed to deal with label imperfection, including robust algorithms by avoid overfitting, filtering mechanism to remove noises and correction mechanism to revise noises. In this paper, we propose an approach based on knowledge graph to perceive and correct the label errors in training data. Experiments on a medical Q&A data set reveal that our knowledge graph based approach can be effective on promoting classification performance and data quality. The results as well show our approach can work in a relatively high noise level and be applied in other data mining tasks demanding deep understanding.

KW - Classification

KW - Data quality

KW - Knowledge graph

KW - Label error

UR - http://www.scopus.com/inward/record.url?scp=84915751769&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-13186-3_32

DO - 10.1007/978-3-319-13186-3_32

M3 - Conference Proceeding

AN - SCOPUS:84915751769

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 345

EP - 356

BT - Trends and Applications in Knowledge Discovery and Data Mining - PAKDD 2014 International Workshops

A2 - Peng, Wen-Chih

A2 - Wang, Haixun

A2 - Zhou, Zhi-Hua

A2 - Ho, Tu Bao

A2 - Tseng, Vincent S.

A2 - Chen, Arbee L.P.

A2 - Bailey, James

PB - Springer Verlag

T2 - International Workshops on Data Mining and Decision Analytics for Public Health, Biologically Inspired Data Mining Techniques, Mobile Data Management, Mining, and Computing on Social Networks, Big Data Science and Engineering on E-Commerce, Cloud Service Discovery, MSMV-MBI, Scalable Dats Analytics, Data Mining and Decision Analytics for Public Health and Wellness, Algorithms for Large-Scale Information Processing in Knowledge Discovery, Data Mining in Social Networks, Data Mining in Biomedical informatics and Healthcare, Pattern Mining and Application of Big Data in conjunction with 18th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2014

Y2 - 13 May 2014 through 16 May 2014

ER -

Liu Y, Li H, Chen Y. Using knowledge graph to handle label imperfection. In Peng WC, Wang H, Zhou ZH, Ho TB, Tseng VS, Chen ALP, Bailey J, editors, Trends and Applications in Knowledge Discovery and Data Mining - PAKDD 2014 International Workshops: DANTH, BDM, MobiSocial, BigEC, CloudSD, MSMV-MBI, SDA, DMDA-Health, ALSIP, SocNet, DMBIH, BigPMA, Revised Selected Papers. Springer Verlag. 2014. p. 345-356. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-13186-3_32

Using knowledge graph to handle label imperfection

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this