Instrumenting the crowd: using implicit behavioral measures to predict task performance
The DOI matches its Crossref record; 21 of the 21 references it cites were found in a registry or an index.
Abstract
Detecting and correcting low quality submissions in crowdsourcing tasks is an important challenge. Prior work has primarily focused on worker outcomes or reputation, using approaches such as agreement across workers or with a gold standard to evaluate quality. We propose an alternative and complementary technique that focuses on the way workers work rather than the products they produce. Our technique captures behavioral traces from online crowd workers and uses them to predict outcome measures such quality, errors, and the likelihood of cheating. We evaluate the effectiveness of the approach across three contexts including classification, generation, and comprehension tasks. The results indicate that we can build predictive models of task performance based on behavioral traces alone, and that these models generalize to related tasks. Finally, we discuss limitations and extensions of the approach.
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
The WEKA data mining software: an update
Hall, M., Frank, E., Holmes, G., Pfahringer, B., Reutemann, P., & Witten, I. H. (2009) · ACM SIGKDD Explorations Newsletter · doi:10.1145/1656274.1656278 · Cited by 17,750 · cited 1×
“…solicit workers on Mechanical Turk, capture their behavior using our prototype, validate that our observations align with their end products, develop a task fingerprint using our observations, and apply the task fingerprint to the evaluation of the task using Weka, a machine learning toolkit [8].”
MethodLabeling images with a computer game
von Ahn, L., & Dabbish, L. (2004) · Proceedings of the SIGCHI Conference on Human Factors in Computing Systems · doi:10.1145/985692.985733 · Cited by 2,204 · cited 1×
“Other approaches include promoting intrinsical motivation [18], splitting larger tasks into small, fault-tolerant subtasks [1, 19], incorporating randomness in cooperative task designs [23], financial manipulation and tweaking outcome”
BackgroundCrowdsourcing user studies with Mechanical Turk
Kittur, A., Chi, E. H., & Suh, B. (2008) · Proceedings of the SIGCHI Conference on Human Factors in Computing Systems · doi:10.1145/1357054.1357127 · Cited by 1,923 · cited 6×
“While task time can be a useful metric for identifying non-compliant workers, as researchers using Mechanical Turk are well aware, task times reported by the system can be dramatically different from the actual time a worker spent on completing a task [15].”
BackgroundMethodResultInfluentialSelf-citationCheap 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 · cited 2×
“Other work calculates relationships between worker answers or identifies erroneous workers using trends to reduce bias [2, 4, 12, 20].”
BackgroundQuality management on Amazon Mechanical Turk
Ipeirotis, P. G., Provost, F., & Wang, J. (2010) · Proceedings of the ACM SIGKDD Workshop on Human Computation · doi:10.1145/1837885.1837906 · Cited by 973 · cited 2×
“Other work calculates relationships between worker answers or identifies erroneous workers using trends to reduce bias [2, 4, 12, 20].”
BackgroundThe state of the art in automating usability evaluation of user interfaces
Ivory, M. Y., & Hearst, M. A. (2001) · ACM Computing Surveys · doi:10.1145/503112.503114 · Cited by 875 · cited 1×
“Ivory and Hearst evaluated a large number of automation systems, concluding that although automated logging is still not widely implemented, it has cost benefits and allows for quick design comparisons [13].”
BackgroundSoylent: a word processor with a crowd inside
Bernstein, M. S., Little, G., Miller, R. C., Hartmann, B., Ackerman, M. S., Karger, D. R., Crowell, D., & Panovich, K. (2010) · Proceedings of the 23nd annual ACM symposium on User interface software and technology · doi:10.1145/1866029.1866078 · Cited by 686 · cited 4×
“A congruent approach is to have workers rate other workers in a verification step [1].”
BackgroundMethodResultInfluentialProceedings of the 2008 Conference on Empirical Methods in Natural Language Processing
Lapata, M., & Ng, H. T. (2008) · Cited by 567
Extracting usability information from user interface events
Hilbert, D. M., & Redmiles, D. F. (2000) · ACM Computing Surveys · doi:10.1145/371578.371593 · Cited by 477 · cited 1×
“HCI researchers have spent considerable effort on capturing and analyzing event logs, primarily for the purposes of evaluating usability [9].”
BackgroundFinancial incentives and the "performance of crowds"
Mason, W., & Watts, D. J. (2010) · ACM SIGKDD Explorations Newsletter · doi:10.1145/1809400.1809422 · Cited by 463 · cited 1×
“Other approaches include promoting intrinsical motivation [18], splitting larger tasks into small, fault-tolerant sub-tasks [1, 19], incorporating randomness in cooperative task designs [23], financial manipulation and tweaking outcome measures [10, 16].”
Method
Cited by
Kittur, A., Nickerson, J. V., Bernstein, M., Gerber, E., Shaw, A., Zimmerman, J., Lease, M., & Horton, J. (2013) · Proceedings of the 2013 conference on Computer supported cooperative work · doi:10.1145/2441776.2441923 · Cited by 1,034
Self-citationDaniel, F., Kucherbaev, P., Cappiello, C., Benatallah, B., & Allahbakhsh, M. (2019) · ACM Computing Surveys · doi:10.1145/3148148 · Cited by 313
Unsupervised Clickstream Clustering for User Behavior Analysis
Wang, G., Zhang, X., Tang, S., Zheng, H., & Zhao, B. Y. (2016) · Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems · doi:10.1145/2858036.2858107 · Cited by 269
Street-Level Algorithms: A Theory at the Gaps Between Policy and Decisions
Alkhatib, A., & Bernstein, M. (2019) · Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3290605.3300760 · Cited by 188
Examining Crowd Work and Gig Work Through The Historical Lens of Piecework
Alkhatib, A., Bernstein, M. S., & Levi, M. (2017) · Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3025453.3025974 · Cited by 148
Model Positionality and Computational Reflexivity: Promoting Reflexivity in Data Science
Cambo, S. A., & Gergle, D. (2022) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3491102.3501998 · Cited by 49
“While similar to the concept of task fngerprinting described in [76], this process aims to be specifc to the task and to make valid comparisons between annotators even”
MethodEvaluating the Quality of Learning Resources: A Learnersourcing Approach
Abdi, S., Khosravi, H., Sadiq, S., & Demartini, G. (2021) · IEEE Transactions on Learning Technologies · doi:10.1109/tlt.2021.3058644 · Cited by 40
Metadata
- Authors
- Rzeszotarski, J. M., & Kittur, A.
- Venue
- Proceedings of the 24th annual ACM symposium on User interface software and technology
- Publisher
- ACM
- Published
- 2011-10-16
- Pages
- 13-22
- OpenAlex
- W2124994029
- Cited by
- 164
- References
- 21
- FWCI
- 8.85
- Type
- conference paper
- Language
- en
- Open access
- closed
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.