Embedding Confidence to Enhance Trust in AI Document Entity Extraction
Bibliographic record
Abstract
Large Language Models (LLMs) are rapidly transforming document understanding by enabling automated extraction of structured data from unstructured sources like resumes and transcripts. Despite high accuracy and efficiency gains, LLMs suffer from a critical limitation: they may hallucinate plausible but incorrect outputs yet provide no explicit confidence measure, undermining reliability in high-stakes domains such as admissions and hiring. This paper presents a practical verification technique for LLM-based entity extraction pipelines, leveraging embedding vector classification to estimate the confidence of each output. We conducted a comparative study of confidence estimation techniques, including LLM self-critique and embeddingbased approaches, on a public resume dataset with synthesized extraction errors. Our findings indicate that embedding-based verification more accurately distinguishes correct from erroneous extractions (F1-score of 0.98), enabling selective flagging of low-confidence fields for human-in-the-loop review. This work advances the adoption of LLMs in sensitive document workflows by providing a scalable, reliable confidence framework.
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 | 0.013 | 0.155 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.005 | 0.003 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.005 | 0.010 |
| Open science | 0.002 | 0.005 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 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".