Metrics

Generative artificial intelligence (AI) in peer review raises questions of permission, human judgment, and fair participation. This normative and conceptual analysis proposes an author-authorised, journal-authorised, version-synchronised public-preprint pathway for bounded external-AI assistance. It aims to broaden opportunities for defensible criticism while maintaining rigorous evaluation and limiting governance to necessary safeguards. Eight cumulative conditions connect voluntary, unanimous author opt-in, journal permission, a rights-cleared public package, round-specific synchronisation, service eligibility, disclosure, human accountability and redress, and exclusion of non-public editorial information. Synchronisation supplies complete, same-version context for checking consequential criticisms without requiring full-manuscript input for every task. An optional open pre-review pilot permits contributions before formal reviewer appointment while preserving invited review. Comparisons separate public access, participation, and AI provision across confidentiality, retention and training, model change, auditability, burden, equity, and complementary questioning. Approved external assistance may support contributors without journal-provided internal-AI accounts; its value depends on voluntary uptake, usable support, contribution quality, and total burdens relative to feasible alternatives. Two-level eligibility checks, proportionate records, staged publication, and independent redress make the proposal inspectable. Effects remain untested. The result is a design for comparative evaluation, not a validated standard, a claim of superiority over internal assistance, or permission to disregard existing policies.

The increasing complexity of evaluation in the higher education setting, along with the growing availability of diverse data sources, has emphasized the need for more transparent and interpretable assessment processes and results. Current evaluation systems frequently depend on fragmented indicators and black-box procedures, hindering evidence-based decision-making. To address these challenges, this study proposes a framework that integrates semantic web technologies, visual analytics and multiple-criteria decision analysis. An ontology is developed to capture evaluation dimensions, criteria, indicators and relationships, enabling interoperability among heterogeneous data sources, including bibliometric databases, institutional repositories, and open science infrastructures. The framework further incorporates a multicriteria method to support transparent weighting and aggregation of evaluation criteria, as well as interactive visual analytics mechanisms designed according to cognition-informed principles. It was applied within a higher education evaluation scenario to support the integration, analysis, and interpretation of multidimensional assessment data. The framework was evaluated using a dataset comprising seven (7) evaluation entities, covering fifty-two (52) evaluation criteria across five (5) dimensions, a cross-cutting dimension and the overall ranking. The ontology-based ELECTRE III analysis involved twenty-one (21) pairwise comparisons and produced seven (7) final ranking positions, with no initial incomparable pairs requiring resolution. At the dimension level, 30 ranking positions were obtained from 105 pairwise comparisons, resulting in five (5) incomparable pairs, four (4) of which were resolved through the proposed SDR mechanism. These results demonstrate the framework’s ability to differentiate the evaluated entities at both the overall and dimension levels while providing an explicit mechanism for resolving ranking incomparabilities. The application aims to facilitate the integration of heterogeneous evaluation data sources, preserve an auditable decision trail, improve the transparency of evaluation procedures, and enhance the interpretation of complex evaluation outcomes. The proposed framework contributes to explainable academic evaluation by combining semantic modeling, decision support, and visual analytics, supporting more informed and transparent institutional decision-making.

In deciding whether to accept or reject one theory rather than another, scientists often look at the evidence that agrees or disagrees with each theory. We formalize this method of comparison in terms of a 2-by-2 matrix where rows are the theories being compared and the columns the extent of agreement or disagreement with evidence. Illustrating this approach with two examples from the history of science, we show under what conditions scientific decisions to support a theory conform with Bayes’ theorem. For example, when a scientist decides that the theory with the most empirical support, or the fewest empirical failures, is the correct one, are they conforming to the rules of Bayesian confirmation? We show by a random simulation that often only a single comparison of agreements or disagreements, successes or failures, is consistent with Bayes’ theorem. This means that many scientists’ decisions to support or reject theories are consistent with Bayes’ theorem without their being aware of it.