Exploring Text Classification Systems for Automatically Coding Historical Occupations and Causes of Death
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
ObjectivesText classification models can be used to automatically categorize occupations and causes of death within historical documents. It is important to classify/code these categories as different words or textual descriptions could refer to the same occupation or cause of death. Given the many historical documents that are becoming available for research, accurate classification systems can be valuable resources. ApproachWe explore different text classification techniques, from traditional machine learning to deep learning, and investigate methodologies that transform occupations and causes of death into a vectorial space and use these representations as features to train text classification systems. Our data come from IPUMS USA/International, and SCADR. ResultsHistorians have coded occupations and causes of death for some census collections (e.g., US, Canada), but not yet for others (e.g., Scotland). We train and evaluate our classification systems using data from the US and Canada and then deploy it on data from Scotland. We quantitatively measure the performance of the classification systems for historical documents that have codes available. Additionally, once we deploy the model to data that does not yet have codes, we qualitatively evaluate our results by engaging with historians working on those data. We report and discuss these results to understand where the models are performing well and where the models are underperforming. ConclusionsResults suggest that there is value in building and deploying these classification models. We recommend the use of such models in conjunction with engaging with domain experts.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.003 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.003 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it