Garbage in, garbage out? Do machine learning application papers in social computing report where human-labeled training data comes from?¶
Geiger, R. S., Yu, K., Yang, Y., Dai, M., Qiu, J., Tang, R., & Huang, J. (2020). Garbage in, garbage out? Do machine learning application papers in social computing report where human-labeled training data comes from?. Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency (FAT, 325-336.
Cited by¶
1 citation across 1 artifact.
Each citation links to the sentence it supports in the citing article.
Primes¶
- Native-Category Flattening
- The asymmetry of recoverability bites hard and expensively — a model trained on these labels can never learn the mid-purchase-versus-overnight distinction, because it was destroyed before training; recovering it requires re-annotating from raw text, if the raw text was even retained.
This sourceDocuments how fixed annotation label sets applied to data carrying finer native distinctions constrain every downstream model and resist after-the-fact recovery.
- The asymmetry of recoverability bites hard and expensively — a model trained on these labels can never learn the mid-purchase-versus-overnight distinction, because it was destroyed before training; recovering it requires re-annotating from raw text, if the raw text was even retained.
Verification¶
Does it exist? Confirmed. This work's DOI resolves to a registered record, which fixes its identity. That is all it fixes.
Does it back the claim? Not recorded. The single citation of this work carries no recorded support check.
Support is checked per citation rather than per work — the same source can be cited soundly in one article and wrongly in another. Per-citation recording began recently, so a citation with no recorded check is a gap in the record rather than evidence it went unchecked.
See how references were verified.
Registry ID ref:dd0c006c9237 · see in the full table