New paper on OSDG 2.0 – a multilingual tool for classifying text data to UN SDGs

November 22, 2022

The OSDG project team has recently published a paper that presents OSDG 2.0 – a new and improved version of the existing SDG classification tool.

Publicly re-launched in May 2022, the tool includes new features, such as classification of PDF files, aside from the initial compatibility with free text input, and support for 15 languages. OSDG users can now classify and analyse content with respect to SDGs in Arabic, Danish, Dutch, Finnish, French, German, Italian, Korean, Polish, Portuguese, Russian, Spanish, Swedish, and Turkish.

The new publication, entitled “OSDG 2.0: a multilingual tool for classifying text data by UN Sustainable Development Goals (SDGs)” is readily available via the open-access repository arXiv.

About the paper

The publication explains the key changes in the classification methodology, first defined in 2020. Newly integrated Machine Learning (ML) models help to uncover the general thematic areas of the SDGs before verifying the SDG labels through ontology/keyword mapping. This results in more accurate and reliable SDG labels.

The paper also details the key milestones of the OSDG Community platform, a global citizen science exercise that contributed to training the OSDG ML models. With over 2,100 volunteers from more than 130 countries, the exercise became a connecting point for SDG researchers, subject matter experts and sustainability advocates, alongside hundreds of UN Volunteers.

The team explains the rationale behind the publicly shared citizen science outputs – the OSDG Community Dataset. Opening up quality SDG datasets is a way for the OSDG team to contribute to the SDG research. With over 3,000 downloads, the quarterly datasets have surely become a trusted asset to researchers in developing their own ontology-based or machine learning approaches for SDG classification. The team also presents examples of existing OSDG applications to encourage continued use of the tool. The paper highlights the existing applications of OSDG’s API, granted free of charge to academic research and scientific communities that aim to produce research papers. The team also celebrates existing creative applications of SDG mapping, as displayed on the citizen science community SciStarter, research repository SwedishScience, and SDG-relevant job vacancy portal SDG Careers. Finally, the team demonstrates several use cases of bespoke research, such as assessing the teaching contributions to SDGs at University College London.

What’s next?

The OSDG team is continuously working to further improve the tool and to uncover overlooked linkages between research areas and the SDGs. The OSDG Community platform exercise remains active and welcomes all SDG experts to contribute to the global consultation.

Are you interested in OSDG methodology and how it can be applied for your data? Contact our team to learn more about our approach to data for universities, companies, and NGOs. The OSDG team also welcomes inquiries for the SDG API, as well as any kind of feedback or suggestions to further improve the tool.