Cognitive psychologists have documented that humans use cognitive heuristics, or mental shortcuts, to make quick decisions while expending less effort. While performing annotation work on crowdsourcing platforms, we hypothesize that such heuristic use among annotators cascades on to data quality and model robustness. In this work, we study cognitive heuristic use in the context of annotating multiple-choice reading comprehension datasets. We propose tracking annotator heuristic traces, where we tangibly measure low-effort annotation strategies that could indicate usage of various cognitive heuristics. We find evidence that annotators might be using multiple such heuristics, based on correlations with a battery of psychological tests. Importantly, heuristic use among annotators determines data quality along several dimensions: (1) known biased models, such as partial input models, more easily solve examples authoredby annotators that rate highly on heuristic use, (2) models trained on annotators scoring highly on heuristic use don’t generalize as well, and (3) heuristic-seeking annotators tend to create qualitatively less challenging examples. Our findings suggest that tracking heuristic usage among annotators can potentially help with collecting challenging datasets and diagnosing model biases.
@inproceedings{malaviya-etal-2022-cascading, title = "Cascading Biases: Investigating the Effect of Heuristic Annotation Strategies on Data and Models", author = "Malaviya, Chaitanya and Bhatia, Sudeep and Yatskar, Mark", editor = "Goldberg, Yoav and Kozareva, Zornitsa and Zhang, Yue", booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing", month = dec, year = "2022", address = "Abu Dhabi, United Arab Emirates", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2022.emnlp-main.438/", doi = "10.18653/v1/2022.emnlp-main.438", pages = "6525--6540", abstract = "Cognitive psychologists have documented that humans use cognitive heuristics, or mental shortcuts, to make quick decisions while expending less effort. While performing annotation work on crowdsourcing platforms, we hypothesize that such heuristic use among annotators cascades on to data quality and model robustness. In this work, we study cognitive heuristic use in the context of annotating multiple-choice reading comprehension datasets. We propose tracking annotator heuristic traces, where we tangibly measure low-effort annotation strategies that could indicate usage of various cognitive heuristics. We find evidence that annotators might be using multiple such heuristics, based on correlations with a battery of psychological tests. Importantly, heuristic use among annotators determines data quality along several dimensions: (1) known biased models, such as partial input models, more easily solve examples authoredby annotators that rate highly on heuristic use, (2) models trained on annotators scoring highly on heuristic use don`t generalize as well, and (3) heuristic-seeking annotators tend to create qualitatively less challenging examples. Our findings suggest that tracking heuristic usage among annotators can potentially help with collecting challenging datasets and diagnosing model biases."}
<?xml version="1.0" encoding="UTF-8"?><modsCollection xmlns="http://www.loc.gov/mods/v3"><mods ID="malaviya-etal-2022-cascading"> <titleInfo> <title>Cascading Biases: Investigating the Effect of Heuristic Annotation Strategies on Data and Models</title> </titleInfo> <name type="personal"> <namePart type="given">Chaitanya</namePart> <namePart type="family">Malaviya</namePart> <role> <roleTerm authority="marcrelator" type="text">author</roleTerm> </role> </name> <name type="personal"> <namePart type="given">Sudeep</namePart> <namePart type="family">Bhatia</namePart> <role> <roleTerm authority="marcrelator" type="text">author</roleTerm> </role> </name> <name type="personal"> <namePart type="given">Mark</namePart> <namePart type="family">Yatskar</namePart> <role> <roleTerm authority="marcrelator" type="text">author</roleTerm> </role> </name> <originInfo> <dateIssued>2022-12</dateIssued> </originInfo> <typeOfResource>text</typeOfResource> <relatedItem type="host"> <titleInfo> <title>Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing</title> </titleInfo> <name type="personal"> <namePart type="given">Yoav</namePart> <namePart type="family">Goldberg</namePart> <role> <roleTerm authority="marcrelator" type="text">editor</roleTerm> </role> </name> <name type="personal"> <namePart type="given">Zornitsa</namePart> <namePart type="family">Kozareva</namePart> <role> <roleTerm authority="marcrelator" type="text">editor</roleTerm> </role> </name> <name type="personal"> <namePart type="given">Yue</namePart> <namePart type="family">Zhang</namePart> <role> <roleTerm authority="marcrelator" type="text">editor</roleTerm> </role> </name> <originInfo> <publisher>Association for Computational Linguistics</publisher> <place> <placeTerm type="text">Abu Dhabi, United Arab Emirates</placeTerm> </place> </originInfo> <genre authority="marcgt">conference publication</genre> </relatedItem> <abstract>Cognitive psychologists have documented that humans use cognitive heuristics, or mental shortcuts, to make quick decisions while expending less effort. While performing annotation work on crowdsourcing platforms, we hypothesize that such heuristic use among annotators cascades on to data quality and model robustness. In this work, we study cognitive heuristic use in the context of annotating multiple-choice reading comprehension datasets. We propose tracking annotator heuristic traces, where we tangibly measure low-effort annotation strategies that could indicate usage of various cognitive heuristics. We find evidence that annotators might be using multiple such heuristics, based on correlations with a battery of psychological tests. Importantly, heuristic use among annotators determines data quality along several dimensions: (1) known biased models, such as partial input models, more easily solve examples authoredby annotators that rate highly on heuristic use, (2) models trained on annotators scoring highly on heuristic use don‘t generalize as well, and (3) heuristic-seeking annotators tend to create qualitatively less challenging examples. Our findings suggest that tracking heuristic usage among annotators can potentially help with collecting challenging datasets and diagnosing model biases.</abstract> <identifier type="citekey">malaviya-etal-2022-cascading</identifier> <identifier type="doi">10.18653/v1/2022.emnlp-main.438</identifier> <location> <url>https://aclanthology.org/2022.emnlp-main.438/</url> </location> <part> <date>2022-12</date> <extent unit="page"> <start>6525</start> <end>6540</end> </extent> </part></mods></modsCollection>
%0 Conference Proceedings%T Cascading Biases: Investigating the Effect of Heuristic Annotation Strategies on Data and Models%A Malaviya, Chaitanya%A Bhatia, Sudeep%A Yatskar, Mark%Y Goldberg, Yoav%Y Kozareva, Zornitsa%Y Zhang, Yue%S Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing%D 2022%8 December%I Association for Computational Linguistics%C Abu Dhabi, United Arab Emirates%F malaviya-etal-2022-cascading%X Cognitive psychologists have documented that humans use cognitive heuristics, or mental shortcuts, to make quick decisions while expending less effort. While performing annotation work on crowdsourcing platforms, we hypothesize that such heuristic use among annotators cascades on to data quality and model robustness. In this work, we study cognitive heuristic use in the context of annotating multiple-choice reading comprehension datasets. We propose tracking annotator heuristic traces, where we tangibly measure low-effort annotation strategies that could indicate usage of various cognitive heuristics. We find evidence that annotators might be using multiple such heuristics, based on correlations with a battery of psychological tests. Importantly, heuristic use among annotators determines data quality along several dimensions: (1) known biased models, such as partial input models, more easily solve examples authoredby annotators that rate highly on heuristic use, (2) models trained on annotators scoring highly on heuristic use don‘t generalize as well, and (3) heuristic-seeking annotators tend to create qualitatively less challenging examples. Our findings suggest that tracking heuristic usage among annotators can potentially help with collecting challenging datasets and diagnosing model biases.%R 10.18653/v1/2022.emnlp-main.438%U https://aclanthology.org/2022.emnlp-main.438/%U https://doi.org/10.18653/v1/2022.emnlp-main.438%P 6525-6540
[Cascading Biases: Investigating the Effect of Heuristic Annotation Strategies on Data and Models](https://aclanthology.org/2022.emnlp-main.438/) (Malaviya et al., EMNLP 2022)