Bridging Ontologies and Computational Workflows: A Framework for Semantic Enrichment and Reproducibility

Computational Science and Engineering (CSE) increasingly relies on complex, data-intensive workflows spanning simulations, and data processing pipelines. With the increasing intricacy of computations and parameterized simulations, the need for reliable, automated frameworks to reproduce computational tasks across platforms has become critical. Computational workflows describe the complex multi-step methods that are used for data collection, data preparation, analytics, predictive modeling, and simulation. These workflows often involve interdependent tasks, data sources, and heterogeneous environments. To manage such complexity and enhance reproducibility, workflow tools are often integrated with semantic technologies. As a part of the MaRDI consortium \cite{mardi} on research data management (RDM) in mathematical sciences, we showcase new features in MaRDIFlow \cite{veluvali2024mardiflow}, a computational and metadata-driven workflow tool, developed for realization and documentation of computer-based experiments. Here, the workflow components are characterized through their input-output relation such that the associate meta-data is used interchangeably and redundantly. For that, we describe suitable meta data and a low level language for the descriptions of general CSE workflows at varying abstraction levels. Importantly, we view redundancy in representing models, code, and data as a strength: it ensures compatibility during execution, and supports reproducibility - the key pillars of replicable and reusable research. In this work, to further enhance the semantic interoperability of our RDM tool, we integrate Voc4Cat \cite{voc4cat}, a domain-specific ontology and SKOS (Simple Knowledge Organization System) vocabulary developed by the NFDI4Cat consortium \cite{nfdi4cat}. Voc4Cat serves as an open, community-driven resource designed to promote the standardization and consistent usage of vocabulary, concepts and definitions in catalysis related sciences. Besides, when linked with MaRDIFlow, Voc4Cat serves as a backbone for annotating and defining metadata dependencies, relationships and supporting workflow descriptions. \begin{figure}[htbp] \centering \input{workflow_components.tikz} \caption{Schematic representation of MaRDIFlow framework integration with Voc4Cat. Each component is defined by its input-output relationship and linked via semantically enriched metadata. Our computational workflow framework supports interchangeable workflow components, integration of Voc4Cat ontology, and documentation scheme for FAIR computational experiments.} \label{fig:workflow_components} \end{figure} A core component of the Voc4Cat ontology integration is via RESTful APIs, enabling real-time querying and metadata validation during workflow construction and execution. Using API endpoints, MaRDIFlow dynamically fetches vocabulary terms, definitions, and relationships, which are then embedded into the workflow metadata from the Voc4Cat database. This ensures semantic consistency across workflow components such that numerical models, input/output data, and mathematical models are accurately represented and aligned with domain-specific standards As a working example within the MaRDIFlow framework, we present a methanization reactor model \cite{bremer2021non}, wherein, we examine the reactor model through a set of nonlinear partial differential equations (PDEs) for mass and energy balances. This catalysis-based use case demonstrates how MaRDIFlow, integrated with Voc4Cat, leverages knowledge graphs to represent and query relationships across multi-layered workflow components. Going forward, we aim to extend MaRDIFlow to interface with additional ontologies within NFDI (Nationale Forschungsdateninfrastruktur), promoting broader interoperability across disciplines and contribute to NFDI's mission to develop community driven research data infrastructure.