How we are stewarding the Physical Chemistry Properties Data Collection to be the best data collection it can be

Partridge, Matthew; Pearman-Kanza, Samantha; Frey, Jeremy

doi:10.5281/zenodo.17594129

How we are stewarding the Physical Chemistry Properties Data Collection to be the best data collection it can be

Partridge, Matthew; Pearman-Kanza, Samantha; Frey, Jeremy

Publisher: Zenodo
DOI: 10.5281/zenodo.17594129
Repository: zenodo.org
Language: en

Abstract

This poster was presented at the Careers and Skills for Data-driven Research (CaSDaR) Townhall Launch event on the 18th September 2025. Abstract: The Physical Chemistry Properties Data Collection (PChemProp), hosted on the Physical Sciences Data Infrastructure (PSDI), is an evolving data collection designed to curate and share high-quality, machine-readable physical chemistry data. This poster presents the current scope of PChemProp, including datasets covering melting points, solubilities, and other key properties, aggregated from diverse sources. It outlines our systematic auditing processes—focusing on metadata completeness, unit standardisation, provenance tracking, and duplicate reconciliation—to ensure the integrity and FAIRness of every entry. We also highlight new developments, including the transition to a JSON-based MongoDB backend, integration of AI-assisted ingestion workflows, and adoption of tools like ProperSea and M-layer to improve semantic structure and unit handling. Looking ahead, we are expanding community engagement and interoperability through schema alignment with OPTIMADE and future web interfaces. This poster showcases how we are stewarding PChemProp to be a robust, sustainable, and valuable resource for researchers and data-driven modelling.

Full text

Physical Chemistry Properties Data Collection PChProp xxxxxxx Matthew Partridge, Samantha Pearman-Kanza, and Jeremy Frey PSDI, School of Chemistry and Chemical Engineering, University of Southampton, SO17 1BJ Context & Challenge •There are multiple large disparate datasets in physical chemistry •Many contain errors and require careful data auditing •Disparete datasets make it hard to use for multi property modelling •Data is being lost or made inaccessible Impact •The data ingest auditing ensures all the data included is of high quality •Opens up the data for modelling and electronic accessibility •Ensures important datasets are easily available to all across the Physical Chemistry domain •AI and ML ready data collections Future •New datasets have been provided to us •This will require careful integration and ongoing auditing •Will need data stewardship to ensure continued quality The Physical Sciences Data Infrastructure is funded through EPSRC Digital Research Infrastructure Funding – Grants EP/X032701/1, EP/X032663/1 and EP/W032252/1. Acknowledgements: We would like to acknowledge and thank all the people involved in the PSDI Statement of Need, Pilot, Phase 1 and Phase 2. Explore more on our website: www.psdi.ac.uk Find us on social media — just search “PSDI”