Anaphora Resolution in Software Requirements Engineering: A Comparison of Generative NLP Pipelines and Encoder-Based Models
Bibliographic record
Abstract
In the field of requirements engineering (RE), anaphoric ambiguity can negatively impact the quality of requirements and could even threaten the success of a project. If different stakeholders like testers or customers interpret software requirements differently, the system might fail to pass the customer validation stage. On the other hand, a robust anaphora resolution model clarifies the writing process of requirements by accurately indicating the pronoun references. In this study, we exploited the power of generative NLP pipelines and compared their performance with the extractive Question Answering (or sequence labeling) technique. We conducted extensive numerical experiments including text-to-text pipelines and compared them with encoder-based models on two public requirements datasets. Our experiments revealed that a sufficiently large T5 model can yield better results than encoder-based models. We've utilized methods such as Lora to effectively address the complexity of training large language models. Our study indicated that the generative approach outperforms classification-based models for anaphora resolution tasks in Software Requirement texts.
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 | 0.006 | 0.014 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.004 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| 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".