Dealing with Disagreements: Looking Beyond the Majority Vote in Subjective Annotations
Open accessThe DOI matches its Crossref record; 24 of the 24 references it cites were found in a registry or an index.
Abstract
Majority voting and averaging are common approaches used to resolve annotator disagreements and derive single ground truth labels from multiple annotations. However, annotators may systematically disagree with one another, often reflecting their individual biases and values, especially in the case of subjective tasks such as detecting affect, aggression, and hate speech. Annotator disagreements may capture important nuances in such tasks that are often ignored while aggregating annotations to a single ground truth. In order to address this, we investigate the efficacy of multi-annotator models. In particular, our multi-task based approach treats predicting each annotators’ judgements as separate subtasks, while sharing a common learned representation of the task. We show that this approach yields same or better performance than aggregating labels in the data prior to training across seven different binary classification tasks. Our approach also provides a way to estimate uncertainty in predictions, which we demonstrate better correlate with annotation disagreements than traditional methods. Being able to model uncertainty is especially useful in deployment scenarios where knowing when not to make a prediction is important.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325
An argument for basic emotions
Ekman, P. (1992) · Cognition and Emotion · doi:10.1080/02699939208411068 · Cited by 9,646 · cited 3×
“While several datasets (Strapparava and Mihalcea, 2007; Buechel and Hahn, 2017) include a small set of labels representing the six Ekman emotions (Ek-man, 1992) — anger , disgust , fear , joy , sadness , and surprise ), or bipolar dimensions of affect — arousal and valence (Russell, 2003), others such as Demszky et al…”
MethodTransformers: State-of-the-Art Natural Language Processing
Wolf, T., Debut, L., Sanh, V., Chaumond, J., Delangue, C., Moi, A., Cistac, P., Rault, T., Louf, R., Funtowicz, M., et al. (2020) · Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations · doi:10.18653/v1/2020.emnlp-demos.6 · Cited by 8,741
Dropout as a bayesian approximation: Representing model uncertainty in deep learning
Gal, Y., & Ghahramani, Z. (2015) · arXiv · doi:10.48550/arxiv.1506.02142 · Cited by 6,471
Core affect and the psychological construction of emotion.
Russell, J. A. (2003) · Psychological Review · doi:10.1037/0033-295x.110.1.145 · Cited by 6,113 · cited 2×
“…the six Ekman emotions (Ek-man, 1992) — anger , disgust , fear , joy , sadness , and surprise ), or bipolar dimensions of affect — arousal and valence (Russell, 2003), others such as Demszky et al. (2020) and Crowdflower (2016) include a wider range of emotion labels according to the Plutchik…”
BackgroundAutomated Hate Speech Detection and the Problem of Offensive Language
Davidson, T., Warmsley, D., Macy, M., & Weber, I. (2017) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v11i1.14955 · Cited by 2,610
Cheap and fast---but is it good?: evaluating non-expert annotations for natural language tasks
Snow, R., O'Connor, B., Jurafsky, D., & Ng, A. Y. (2008) · Proceedings of the Conference on Empirical Methods in Natural Language Processing - EMNLP '08 · doi:10.3115/1613715.1613751 · Cited by 1,887
Advances in natural language processing
Hirschberg, J., & Manning, C. D. (2015) · Science · doi:10.1126/science.aaa8685 · Cited by 1,712
Maximum Likelihood Estimation of Observer Error-Rates Using the EM Algorithm
Dawid, A. P., & Skene, A. M. (1979) · Applied Statistics · doi:10.2307/2346806 · Cited by 1,692 · cited 1×
“In early work, Dawid and Skene (1979) used the EM algorithm to obtain maximum likelihood estimates of the “true” label to account for annotator errors.”
MethodGoEmotions: A Dataset of Fine-Grained Emotions
Demszky, D., Movshovitz-Attias, D., Ko, J., Cowen, A., Nemade, G., & Ravi, S. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.372 · Cited by 873
GoEmotions: A Dataset of Fine-Grained Emotions
Demszky, D., Movshovitz-Attias, D., Ko, J., Cowen, A., Nemade, G., & Ravi, S. (2020) · arXiv · doi:10.48550/arxiv.2005.00547 · Cited by 873
The Risk of Racial Bias in Hate Speech Detection
Sap, M., Card, D., Gabriel, S., Choi, Y., & Smith, N. A. (2019) · Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/p19-1163 · Cited by 817 · cited 1×
“Similarly these models are shown to overestimate the prevalence of toxicity in African American Vernacular English (Sap et al., 2019; Davidson et al., 2019; Zhou et al., 2021).”
BackgroundMeasuring and Mitigating Unintended Bias in Text Classification
Dixon, L., Li, J., Sorensen, J., Thain, N., & Vasserman, L. (2018) · Proceedings of the 2018 AAAI/ACM Conference on AI, Ethics, and Society · doi:10.1145/3278721.3278729 · Cited by 674 · cited 1×
“For instance, toxicity prediction models were shown to have biases towards mentions of certain identity terms (Dixon et al., 2018), specific named entities (Prabhakaran et al., 2019), and disabilities (Hutchinson et al., 2020).”
BackgroundRacial Bias in Hate Speech and Abusive Language Detection Datasets
Davidson, T., Bhattacharya, D., & Weber, I. (2019) · Proceedings of the Third Workshop on Abusive Language Online · doi:10.18653/v1/w19-3504 · Cited by 414 · cited 1×
“Similarly these models are shown to overestimate the prevalence of toxicity in African American Vernacular English (Sap et al., 2019; Davidson et al., 2019; Zhou et al., 2021).”
BackgroundA BERT-Based Transfer Learning Approach for Hate Speech Detection in Online Social Media
Mozafari, M., Farahbakhsh, R., & Crespi, N. (2020) · Studies in Computational Intelligence · doi:10.1007/978-3-030-36687-2_77 · Cited by 396
Geva, M., Goldberg, Y., & Berant, J. (2019) · Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) · doi:10.18653/v1/d19-1107 · Cited by 284 · cited 1×
“Researchers have pointed out that in more ob-jective tasks, such as commonsense knowledge or word sense disambiguation, training a model on judgements of a specific set of annotators lack generalizability to annotations generated by new annotators (Geva et al., 2019).”
BackgroundSocial Biases in NLP Models as Barriers for Persons with Disabilities
Hutchinson, B., Prabhakaran, V., Denton, E., Webster, K., Zhong, Y., & Denuyl, S. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.487 · Cited by 274
Self-citationSocial Biases in NLP Models as Barriers for Persons with Disabilities
Hutchinson, B., Prabhakaran, V., Denton, E., Webster, K., Zhong, Y., & Denuyl, S. (2020) · arXiv · doi:10.48550/arxiv.2005.00813 · Cited by 274
Self-citationThe Benefits of a Model of Annotation
Passonneau, R. J., & Carpenter, B. (2014) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00185 · Cited by 227
Learning from the Worst: Dynamically Generated Datasets to Improve Online Hate Detection
Vidgen, B., Thrush, T., Waseem, Z., & Kiela, D. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.132 · Cited by 174
Crossmod: A Cross-Community Learning-based System to Assist Reddit Moderators
Chandrasekharan, E., Gandhi, C., Mustelier, M. W., & Gilbert, E. (2019) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3359276 · Cited by 151 · cited 1×
“For example, the level of uncertainty could help determine when and how moderators take part in a human-in-the-loop content moderation (Chandrasekharan et al., 2019; Liu, 2020).”
BackgroundAddressing age-related bias in sentiment analysis,
Diaz, M., Johnson, I., Lazar, A., Piper, A. M., & Gergle, D. (2018) · Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3173574.3173986 · Cited by 138 · cited 1×
“…moral values, and lived experiences often influence their interpretations, especially in subjective tasks such as identifying political stances (Luo et al., 2020), sentiment (Díaz et al., 2018), and online abuse and hate speech (Cowan and Khatchadourian, 2003; Waseem, 2016; Patton et al., 2019).”
BackgroundSelf-citationPerturbation Sensitivity Analysis to Detect Unintended Model Biases
Prabhakaran, V., Hutchinson, B., & Mitchell, M. (2019) · Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) · doi:10.18653/v1/d19-1578 · Cited by 115
Self-citationThe Disagreement Deconvolution: Bringing Machine Learning Performance Metrics In Line With Reality
Gordon, M. L., Zhou, K., Patel, K., Hashimoto, T., & Bernstein, M. S. (2021) · Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3411764.3445423 · Cited by 101
Cited by
Bias and Fairness in Large Language Models: A Survey
Gallegos, I. O., Rossi, R. A., Barrow, J., Tanjim, M. M., Kim, S., Dernoncourt, F., Yu, T., Zhang, R., & Ahmed, N. K. (2024) · Computational Linguistics · doi:10.1162/coli_a_00524 · Cited by 1,245
ToxiGen: A Large-Scale Machine-Generated Dataset for Adversarial and Implicit Hate Speech Detection
Hartvigsen, T., Gabriel, S., Palangi, H., Sap, M., Ray, D., & Kamar, E. (2022) · Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2022.acl-long.234 · Cited by 279
ToxiGen: A Large-Scale Machine-Generated Dataset for Adversarial and Implicit Hate Speech Detection
Hartvigsen, T., Gabriel, S., Palangi, H., Sap, M., Ray, D., & Kamar, E. (2022) · arXiv · doi:10.48550/arxiv.2203.09509 · Cited by 276
Who Validates the Validators? Aligning LLM-Assisted Evaluation of LLM Outputs with Human Preferences
Shankar, S., Zamfirescu-Pereira, J. D., Hartmann, B., Parameswaran, A., & Arawjo, I. (2024) · Proceedings of the 37th Annual ACM Symposium on User Interface Software and Technology · doi:10.1145/3654777.3676450 · Cited by 151
Feng, S., Park, C. Y., Liu, Y., & Tsvetkov, Y. (2023) · Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2023.acl-long.656 · Cited by 149
Metadata
- Authors
- Davani, A. M., Díaz, M., & Prabhakaran, V.
- Venue
- Transactions of the Association for Computational Linguistics
- Publisher
- MIT Press
- Published
- 2022
- Volume
- 10
- Pages
- 92-110
- ISSN
- 2307-387X (ISSN-L)
- OpenAlex
- W3206428286
- Cited by
- 219
- References
- 24
- FWCI
- 14.30
- Type
- journal article
- Language
- en
- Open access
- diamond
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.