Predicting Verification Methods from Natural Language Requirements
Bibliographic record
Abstract
Abstract – A Verification Cross-Reference Matrix (VCRM) is a table that depicts the verification methods for requirements in a specification. Usually requirement labels are rows, available test methods are columns, and an “X” in a cell indicates usage of a verification method for that requirement. Verification methods include Demonstration, Inspection, Analysis and Test, and sometimes Certification, Similarity and/or Analogy. VCRMs enable acquirers and stakeholders to quickly understand how a product’s requirements will be tested.Maintaining consistency of very large VCRMs can be challenging, and inconsistent verification methods can result in a large set of uncoordinated “spaghetti tests”. Natural language processing algorithms that can identify similarities between requirements offer promise in addressing this challenge.This paper applies and compares compares four natural language processing algorithms to the problem of automatically populating VCRMs from natural language requirements: Naïve Bayesian inference, (b) Nearest Neighbor by weighted Dice similarity, (c) Nearest Neighbor with Latent Semantic Analysis similarity, and (d) an ensemble method combining the first three approaches. The VCRMs used for this study are for slot machine technical requirements derived from gaming regulations from the countries of Australia and New Zealand, the province of Nova Scotia (Canada), the state of Michigan (United States) and recommendations from the International Association of Gaming Regulators (IAGR).
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 | 0.004 | 0.039 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.003 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 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".