Temporal Reasoning in Historical Humanities Data: The Case of Muzio Clementi and the Music Trade

This is a particularly clear paper which addresses a common issue with the modeling of temporal data. The approach is technical enough to cover the complexity of the issues as well as the breadth of the corpus while being easily accessible and understandable.

The paper presents a good variety of use cases while refraining from dabbling in anecdote. Relationships inferred and shown on page 6 (“during” relations) are particularly interesting. It would have been interesting to know more about how those temporal relationships relate to musicological perspectives; indeed, what is presented on page 7 regarding the composition of Clementi’s Op. 7 has potential to be developed more in depth around more concrete examples.

The part “Valuable vs. trivial inferences” is a bit trivial and could be shortened in favor of a more aesthetically informed consideration of his compositional output through the temporal data prism. The temporal reasoning has been explicated with enough clarity in the first pages.

It is overall a strong article with potential for further systemic applications.

In this work, the authors show how to overcome vague dates in historical sources by encoding events about Muzio Clementi in the CIDOC CRM ontology, applying Allen’s interval logic via Notation3 rules, and running the EYE reasoner to infer both valuable and irrelevant temporal relationships. The former can provide interesting insights about interconnected historical events, by filling gaps between them and enriching current knowledge.
The paper is clear and well-written, by undescoring also the current limitations of the approach, as author themsleves declare: “modeling historical events requires making assumptions about event timing, duration, or sequence—and differing assumptions can lead to various outcomes. A single dataset can support multiple valid reconstructions, depending on how uncertainty is addressed and the inferences that are drawn”.
The above assumptions may introduce biases depending on current uncertainty and persona-interpretation, that may vary across different times, contexts and annotators. Sincerely, I have some concerns about the soundness of such inferred relationships, without introducing also a confidence account that must be somehow also quantified. Even if for the case of Muzio Clementi the impact of such arbitrariness can be acceptable, there can be other case-study with higher level of uncertainty which would decrease outcomes soundness. As authors themselves claimed at the end of the paper (as future works), the employing of the new CIDOC CRM temporal relations supports “fuzzy” reasoning about intervals with imprecise boundaries: that would have been more worthwhile.
Still, the paper entirely lacks a proper Related Work section that provides an exhaustive comparison with similar approaches, including those based on machine learning techniques such as GATs, GNNs, and LLM-based methods. The section titled The Current State of the Art merely cites the sources of the technologies employed by the proposed approach and othre assumption (OWA), rather than critically discussing why one should prefer the authors’ method (despite its acknowledged biases) over other state-of-the-art alternatives.

The manuscript was submitted as a full paper and addresses a relevant and non-trivial problem in the area of temporal reasoning for humanities data. Overall, the contribution is very strong. The core intuition is sound and valuable, the methodological approach is rigorous, and the proposed framework shows clear potential for reuse beyond the specific historical and documentary domain examined in the paper.

(1) Originality

The paper presents a highly original contribution. While Allen’s Interval Algebra has been widely discussed in the literature, its systematic application to historically imprecise temporal data, combined with CIDOC CRM modeling and an RDF-based reasoning framework, is both innovative and well motivated.

Particularly noteworthy is the authors’ explicit engagement with the limitations of Allen’s algebra in historical contexts, alongside their discussion of more “fuzzy” temporal algebras and recent CIDOC CRM developments. This reflexive stance strengthens the originality of the work, as it does not merely apply an existing formalism, but critically situates it within an evolving landscape of temporal reasoning approaches.

(2) Significance of the results

The results are significant and convincing. The large-scale inference of temporal relations demonstrates the expressive power of the proposed method and its ability to extract structured temporal knowledge from incomplete and uncertain data.

Importantly, the manuscript goes beyond quantitative output by distinguishing between historically meaningful inferences and logically valid but interpretively trivial ones. This distinction substantially increases the scholarly value of the work and shows a mature understanding of how automated reasoning can support—rather than replace—human interpretation.

The method appears potentially applicable to a wide range of research contexts, including domains not strictly limited to historical or documentary data, wherever temporally imprecise events must be formally modeled and reasoned over.

(3) Quality of writing

The paper is well written, clear, and well structured. The argumentation is coherent throughout, and the methodological choices are carefully justified. Technical sections are detailed without being obscure, and the balance between formal explanation and domain-specific examples is effective.

Given the complexity of the topic, the manuscript succeeds in remaining accessible to readers from both the digital humanities and semantic web communities.

(4) Assessment of data and resources

(A) Organization and documentation
The data resources appear to be well organized, and the presence of explanatory material (including a README file) facilitates understanding of the dataset structure, modeling choices, and assumptions. This significantly lowers the barrier to reuse.

(B) Completeness for replication
The provided resources appear largely sufficient for replication of the reported experiments, including the temporal modeling framework and the reasoning setup. As is often the case with historically curated datasets, full replication may still require domain expertise, but this does not detract from the overall transparency and reproducibility of the work.

(C) Repository choice and long-term accessibility: it could be useful to provide a repository link for data and reasoning scripts

(4) Completeness of data artifacts
Overall, the data artifacts are complete and coherent with the claims made in the paper, and they effectively support the methodological and experimental sections of the manuscript.

This is a very good paper. The intuition behind the work is strong and valuable, the experimentation is rigorous, and the methodological framework is robust and extensible. The manuscript makes a clear contribution to research on temporal reasoning and has the potential to influence future work well beyond the specific historical case study presented.