About the project

In short, OSDG-mapping builds an integrated ontology from the feature sets identified in previous research, and then matches the ontology items to the topics from Microsoft Academic.

Check out our paper on ArXiv OSDG -- Open-Source Approach to Classify Text Data by UN Sustainable Development Goals (SDGs)

OSDG takes relevant text features (such as ontology items, features from machine-learning models or extracted keywords) from the previous research, cleans them and merges them into a comprehensive, constantly-growing OSDG ontology. The ontology items are mapped to the ever-growing list of topics/Fields of Study in the Microsoft Academic Graph (MAG). By doing this, we:

• expand the ontology – acquire more key terms associated with the relevant MAG Topics, natively called Fields of Study (FOS);
• capture more nuanced relationships between individual terms and latent concepts.

OSDG-mapping integrates the existing research into a comprehensive approach, and does so in a way that evades the shortcomings of former individual approaches and duplication of research efforts.

What OSDG-mapping is for?

OSDG-mapping constructs SDG relevant FOS ontology which is an important element of OSDG-tool .

Procedure

Assigned labels from raw data sources are assembled in two steps:

1. Assembling terms AssemblingTerms.py
   Assembles terms from raw_data/0_add/ data sources.

   • Term label conflicts from sources 00_add_validated/ are ignored meaning if term_1 is assigned to SDG_1 by source_1 and to SDG_2 by source_2 → term_1 is assigned to both.
   • Conflicts for term labels from 01_add_generated/ data sources are managed in two ways:
     • If the conflict is between validated and generated term label → generated term label is discarded while validated one remains.
     • If the conflict is between generated & generated → both are discarded.

   → produces InterimTerms.json

   {
       'SDG_1': {
           'term_1': ['source_1', 'source_2', ...],
           'term_2': ['source_1', 'source_3', ...]
           ...
       }
       ...
   }

2. Assembling OSDG Ontology AssemblingOntology.py
   Assembles FOS from InterimTerms.json and 02_add_all_to_all/ data sources.

   • 2.1. Terms from InterimTerms.json are matched to MAG Fields of Study subset FOSMAP.json which contains over 150 thousand fields.
   • 2.2. Matched FOS are added to the final ontology FOS-Ontology.json .
   • 2.3. 02_add_all_to_all/ FOS are added to the final ontology FOS-Ontology.json .
   • 2.4 Final ontology OSDG-Ontology.json is adjusted based on 1_replace/ and 2_remove/ .

   → produces OSDG-Ontology.json

   {
       'SDG_1': ['fos_id_1', 'fos_id_2', ...],
       'SDG_2': ['fos_id_3', 'fos_id_4', ...]
       ...
   }

---

References and inspiration

The list of data sources used in the current version of the OSDG Tool are here. OSDG leverages the data from Microsoft Academic:

1. Sinha, A., Shen, Z., Song, Y., Ma, H., Eide, D., Hsu, B.-J. & Wang, K. (2015). AnOverview of Microsoft Academic Service (MAS) and Applications. Proceedings of the24th International Conference on World Wide Web (p./pp. 243--246), Republic andCanton of Geneva, Switzerland: International World Wide Web Conferences SteeringCommittee. ISBN: 978-1-4503-3473-0. doi:10.1145/2740908.27428398.
2. Wang, K., Shen, Z., Huang, C., Wu, C., Eide, D., Dong, Y., Qian, J., Kanakia, A., Chen,A.C., & Rogahn, R. (2019). A Review of Microsoft Academic Services for Science ofScience Studies. Frontiers in Big Data, 2. doi:10.3389/FDATA.2019.00045