5610131 A STANDARDISED LANGUAGE TAG PROTOCOL AND WORKFLOW FOR DEVELOPING AND MANAGING MULTILINGUAL ONTOLOGIES: SHOWCASING THE FRENCH SICKLE CELL DISEASE ONTOLOGY (SCDO)

Abstract

Background: All Francophone African countries are affected by sickle cell disease (SCD). The collaboratively developed Sickle Cell Disease Ontology (SCDO), which is to our knowledge currently the most comprehensive resource of standardised hierarchical descriptions of knowledge in the SCD domain (Nembaware et al., 2020), is however available only in English. Existing widely used methods for producing consistent translations, such as the back-translation method (Sergiy, 2011) and the more rigorous FACIT translation and linguistic validation methodology (Eremenco et al. 2005), are time consuming and costly (Wołk and Wołk 2017) and in fact have been criticised recently for negatively affecting health research results (Colina et al., 2017). More recently, an alternative functionalist approach has been proposed as a better option for translating text involved in health research (Colina et al., 2017). Aim: We aimed to produce a French translation of the SCDO, as it would promote SCDO application by Francophone clinicians, researchers and other relevant stakeholders, aiding SCDO-facilitated harmonisation of data from Francophone research sites and making patient-centred tracking, diagnostics and therapeutics applications more accessible to Francophone patients. Methods: In order to translate the SCDO into French, we created a customised functionalist approach. We used the language translation software, Babylon, to generate French translations of labels and definitions of SCDO terms and had these French auto-translations manually reviewed by French-speaking colleagues in Tunisia. To manage translation of the SCDO, including future updates of the French translation as well as translations into other languages, we developed a graphical user interface (GUI), coded using Python, and wrote standard operating procedures (SOPs) for using the GUI. To display the French translations, as well as future layperson translations, in the SCDO OWL file, we developed a set of language tags using the World Wide Web Consortium’s IANA Language Subtag Registry and the language tag syntax defined by the Internet Engineering Task Force’s (IETF’s) Best Current Practice (BCP) 47. Results: We have produced 1) the first French translation of the SCDO, 2) an Ontology Translation GUI, 3) SOPs for using the GUI, and 4) a language tag protocol which we propose for use in translated biomedical ontologies. The French label and definition for each SCDO term can now be displayed by clicking a button on the term’s webpage, which is searchable on the SCDO website (https://scdontology.h3abionet.org/). Summary - Conclusion: We describe a novel time-saving and cost-effective workflow, with associated tools, including a standardised protocol of language tags, for the translation of ontologies, and highlight the utility of our method with its successful application in producing the French SCDO. We expect our workflow will be easy to adapt to suit the requirements of other ontology translation projects.