Back to Search Start Over

Harmony: a web-tool for retrospective, multilingual harmonisation of questionnaire items using natural language processing

Authors :
Bettina Moltrecht
Thomas Andrew Wood
Mauricio Scopel Hoffmann
Eoin McElroy
Publication Year :
2023
Publisher :
Center for Open Science, 2023.

Abstract

MotivationRetrospective questionnaire harmonisation allows researchers to pool and analyse information from multiple data sources, thereby increasing reproducibility in science. Currently, harmonisation of questionnaires relies on a multi-step process where items are manually matched based on expert opinion. Harmony, a new natural-language processing tool supports researchers with fast, reproducible and multilingual measurement harmonisation. ImplementationHarmony is a web tool which can be used from any major web-browser by non-technical users.General features Users can upload questionnaire meta-data in text, PDF or excel format. Once uploaded, Harmony extracts the relevant information from the files and presents them to the user. Users choose which items they would like to match. Harmony will use the cosine similarity to find the closest matches between items and present the findings including a percentage match-score. AvailabilityHarmony is open-source (//github.com/harmonydata/harmony) and freely available via //app.harmonydata.org/ under the MIT licence for non-commercial use.

Details

Database :
OpenAIRE
Accession number :
edsair.doi.dedup.....509b3d91f54d33a269afbc41ff886786
Full Text :
https://doi.org/10.31234/osf.io/zvqbf