Bibliographic record
Abstract
In large-scale model-based development, developers periodically need to combine collections of interrelated models. These models may capture different features of a system, describe alternative perspectives on a single feature, or express ways in which different features alter one another's structure or behaviour. We refer to the process of combining a set of interrelated models as "model fusion". A number of factors make model fusion complicated. Models may overlap, in that they refer to the same concepts, but these concepts may be presented differently in each model, and the models may contradict one another. Models may describe independent system components, but the components may interact, potentially causing undesirable side effects. Finally, models may cross-cut, modifying one another in ways that violate their syntactic or semantic properties. In this thesis, we study three instances of the fusion problem for "behavioural models", motivated by real-world applications. The first problem is combining "partial" models of a single feature with the goal of creating a more complete description of that feature. The second problem is maintenance of "variant" specifications of individual features. The goal here is to combine the variants while preserving their points of difference (i.e., variabilities). The third problem is analysis of interactions between models describing "different" features. Specifically, given a set of features, the goal is to construct a composition such that undesirable interactions are absent. We provide an automated tool-supported solution to each of these problems and evaluate our solutions. The main novelties of the techniques presented in this thesis are (1) preservation of semantics during the fusion process, and (2) applicability to large and evolving collections of models. These are made possible by explicit modelling of partiality, variability and regularity in behavioural models, and providing semantic-preserving notions for relating these models.
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 | 0.011 | 0.027 |
| Meta-epidemiology (narrow) | 0.002 | 0.002 |
| Meta-epidemiology (broad) | 0.001 | 0.005 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.002 | 0.003 |
| Scholarly communication | 0.004 | 0.010 |
| Open science | 0.004 | 0.010 |
| Research integrity | 0.003 | 0.005 |
| Insufficient payload (model declined to judge) | 0.009 | 0.003 |
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".