An End-to-End Workflow for Processing Multilingual Stakeholder Workshop Data: A Soil Health Case Study

Vid Podpečan, Bojan Blažica, Fabio Volkmann, Carmen Vazquez Martin, Rachel Creamer, Marko Debeljak

This paper presents an end-to-end workflow for the collection, analysis, and dissemination of multilingual stakeholder workshop data related to soil health. Stakeholder workshops often produce diverse qualitative and ordinal data which is difficult to process consistently and transparently, especially in multilingual settings. The proposed workflow provides clear guidance for collecting, translating, organising, analysing, and reporting data originating from stakeholder workshops. The workflow covers the complete process from data collection, preparation and structured storage to analysis, reporting, and dissemination. It builds upon a range of data analysis methods, including large language models. It is designed to support the analysis of diverse types of data and enables both qualitative exploration and comparison based on derived numerical scores and rankings. We also propose a structured approach based on large language models to topic extraction and topic intensity scoring which allows comparing stakeholder perspectives across workshops and contexts. Finally, a templated reporting process and an interactive online tool support clear and consistent communication of results to stakeholders and other audiences. The proposed workflow is demonstrated in a large European research project involving multiple workshops, stakeholder groups, land-use contexts, and languages. The main contribution of this work is a transparent and adaptable workflow that integrates multilingual data handling, analysis, and reporting into a single framework for stakeholder-based soil health research.