Automated Question Answering for Improved Understanding of Compliance Requirements: A Multi-Document Study
Bibliographic record
Abstract
Software systems are increasingly subject to regulatory compliance. Extracting compliance requirements from regulations is challenging. Ideally, locating compliance-related information in a regulation requires a joint effort from requirements engineers and legal experts, whose availability is limited. However, regulations are typically long documents spanning hundreds of pages, containing legal jargon, applying complicated natural language structures, and including cross-references, thus making their analysis effort-intensive. In this paper, we propose an automated question-answering (QA) approach that assists requirements engineers in finding the legal text passages relevant to compliance requirements. Our approach utilizes large-scale language models fine-tuned for QA, including BERT and three variants. We evaluate our approach on 107 question-answer pairs, manually curated by subject-matter experts, for four different European regulatory documents. Among these documents is the general data protection regulation (GDPR) – a major source for privacy-related requirements. Our empirical results show that, in $\approx 94$% of the cases, our approach finds the text passage containing the answer to a given question among the top five passages that our approach marks as most relevant. Further, our approach successfully demarcates, in the selected passage, the right answer with an average accuracy of $\approx$91%.
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 | 0.012 | 0.049 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.005 | 0.003 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.003 | 0.006 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
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".