Modeling annotator expertise: Learning when everybody knows a bit of something

Yan Yan, Romer Rosales, Glenn Fung, Mark Schmidt, Gerardo Hermosillo, Luca Bogoni, Linda Moy, Jennifer Dy
Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, PMLR 9:932-939, 2010.

Abstract

Supervised learning from multiple labeling sources is an increasingly important problem in machine learning and data mining. This paper develops a probabilistic approach to this problem when annotators may be unreliable (labels are noisy), but also their expertise varies depending on the data they observe (annotators may have knowledge about different parts of the input space). That is, an annotator may not be consistently accurate (or inaccurate) across the task domain. The presented approach produces classification and annotator models that allow us to provide estimates of the true labels and annotator variable expertise. We provide an analysis of the proposed model under various scenarios and show experimentally that annotator expertise can indeed vary in real tasks and that the presented approach provides clear advantages over previously introduced multi-annotator methods, which only consider general annotator characteristics.

Cite this Paper


BibTeX
@InProceedings{pmlr-v9-yan10a, title = {Modeling annotator expertise: Learning when everybody knows a bit of something}, author = {Yan, Yan and Rosales, Romer and Fung, Glenn and Schmidt, Mark and Hermosillo, Gerardo and Bogoni, Luca and Moy, Linda and Dy, Jennifer}, booktitle = {Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics}, pages = {932--939}, year = {2010}, editor = {Teh, Yee Whye and Titterington, Mike}, volume = {9}, series = {Proceedings of Machine Learning Research}, address = {Chia Laguna Resort, Sardinia, Italy}, month = {13--15 May}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v9/yan10a/yan10a.pdf}, url = {https://proceedings.mlr.press/v9/yan10a.html}, abstract = {Supervised learning from multiple labeling sources is an increasingly important problem in machine learning and data mining. This paper develops a probabilistic approach to this problem when annotators may be unreliable (labels are noisy), but also their expertise varies depending on the data they observe (annotators may have knowledge about different parts of the input space). That is, an annotator may not be consistently accurate (or inaccurate) across the task domain. The presented approach produces classification and annotator models that allow us to provide estimates of the true labels and annotator variable expertise. We provide an analysis of the proposed model under various scenarios and show experimentally that annotator expertise can indeed vary in real tasks and that the presented approach provides clear advantages over previously introduced multi-annotator methods, which only consider general annotator characteristics.} }
Endnote
%0 Conference Paper %T Modeling annotator expertise: Learning when everybody knows a bit of something %A Yan Yan %A Romer Rosales %A Glenn Fung %A Mark Schmidt %A Gerardo Hermosillo %A Luca Bogoni %A Linda Moy %A Jennifer Dy %B Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2010 %E Yee Whye Teh %E Mike Titterington %F pmlr-v9-yan10a %I PMLR %P 932--939 %U https://proceedings.mlr.press/v9/yan10a.html %V 9 %X Supervised learning from multiple labeling sources is an increasingly important problem in machine learning and data mining. This paper develops a probabilistic approach to this problem when annotators may be unreliable (labels are noisy), but also their expertise varies depending on the data they observe (annotators may have knowledge about different parts of the input space). That is, an annotator may not be consistently accurate (or inaccurate) across the task domain. The presented approach produces classification and annotator models that allow us to provide estimates of the true labels and annotator variable expertise. We provide an analysis of the proposed model under various scenarios and show experimentally that annotator expertise can indeed vary in real tasks and that the presented approach provides clear advantages over previously introduced multi-annotator methods, which only consider general annotator characteristics.
RIS
TY - CPAPER TI - Modeling annotator expertise: Learning when everybody knows a bit of something AU - Yan Yan AU - Romer Rosales AU - Glenn Fung AU - Mark Schmidt AU - Gerardo Hermosillo AU - Luca Bogoni AU - Linda Moy AU - Jennifer Dy BT - Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics DA - 2010/03/31 ED - Yee Whye Teh ED - Mike Titterington ID - pmlr-v9-yan10a PB - PMLR DP - Proceedings of Machine Learning Research VL - 9 SP - 932 EP - 939 L1 - http://proceedings.mlr.press/v9/yan10a/yan10a.pdf UR - https://proceedings.mlr.press/v9/yan10a.html AB - Supervised learning from multiple labeling sources is an increasingly important problem in machine learning and data mining. This paper develops a probabilistic approach to this problem when annotators may be unreliable (labels are noisy), but also their expertise varies depending on the data they observe (annotators may have knowledge about different parts of the input space). That is, an annotator may not be consistently accurate (or inaccurate) across the task domain. The presented approach produces classification and annotator models that allow us to provide estimates of the true labels and annotator variable expertise. We provide an analysis of the proposed model under various scenarios and show experimentally that annotator expertise can indeed vary in real tasks and that the presented approach provides clear advantages over previously introduced multi-annotator methods, which only consider general annotator characteristics. ER -
APA
Yan, Y., Rosales, R., Fung, G., Schmidt, M., Hermosillo, G., Bogoni, L., Moy, L. & Dy, J.. (2010). Modeling annotator expertise: Learning when everybody knows a bit of something. Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 9:932-939 Available from https://proceedings.mlr.press/v9/yan10a.html.

Related Material