Reasoning about Consistency in Model Merging.
Bibliographic record
Abstract
Models undergo a variety of transformations throughout development. One of the key transformations is merge, used when developers need to combine a set of models with respect to the overlaps between them. A major question about model transformations in general, and merge in particular, is what consistency properties are preserved across the transformations and what consistency properties may need to be re-checked (and if necessary, re-established) over the result. In previous work [18], we developed a technique based on category-theoretic colimits for merging sets of inter-related models. The use of category theory leads to the preservation of the algebraic structure of the source models in the merge; however, this does not directly provide a characterization of the (in)consistency properties that carry over from the source models to the result, because consistency properties are predominantly expressed as logical formulas. Hence, an investigation of the connections between the “algebraic ” and “logical ” properties of model merging became necessary. In this paper, we undertake such an investigation and use techniques from finite model theory [9] to show that the use of colimits indeed leads to the preservation of certain logical properties. Our results have implications beyond the merge framework in [18] and are potentially useful for the broad range of techniques in the graph transformation and algebraic specification literature that use colimits as the basis for model manipulations.
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 | 0.037 | 0.133 |
| Meta-epidemiology (narrow) | 0.002 | 0.003 |
| Meta-epidemiology (broad) | 0.002 | 0.006 |
| Bibliometrics | 0.006 | 0.005 |
| Science and technology studies | 0.006 | 0.014 |
| Scholarly communication | 0.010 | 0.039 |
| Open science | 0.007 | 0.015 |
| Research integrity | 0.005 | 0.009 |
| Insufficient payload (model declined to judge) | 0.004 | 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".