Leveraging AI to Investigate Child Maltreatment Text Narratives: Promising Benefits and Addressable Risks
Bibliographic record
Abstract
Unlabelled: The trove of information contained in child maltreatment narratives represents an opportunity to strengthen the evidence base for policy reform in this area, yet it remains underutilized by researchers and policy makers. Current research into child maltreatment often involves the use of qualitative methodologies or structured survey data that are either too broad or not representative, thereby limiting the development of effective policy responses and intervention strategies. Artificial intelligence (AI) approaches such as large language models (AI models that understand and generate language) can analyze large volumes of child maltreatment narratives by extracting population-level insights on factors of interest such as mental health and treatment needs. However, when applying such methods, it is useful to have a framework on which to base approaches to the data. We propose a seven step framework: (1) data governance; (2) researcher vetting; (3) data deidentification; (4) data access; (5) feasibility testing of baseline methods; (6) large-scale implementation of black box algorithms; and (7) domain expert result validation for such exercises to ensure careful execution and limit the risk of privacy and security breaches, bias, and unreliable conclusions.
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.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.057 | 0.191 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.008 | 0.006 |
| Science and technology studies | 0.002 | 0.003 |
| Scholarly communication | 0.008 | 0.011 |
| Open science | 0.003 | 0.006 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.007 | 0.002 |
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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".