<aside> 📌 AI assistance for partner matching! ✨

</aside>

Previously, when synchronizing Knowledge Products from CGSpace with the PRMS Reporting Tool, there was no standardized data for institutions (author affiliations) that we could store in the system's database. Consequently, at the end of each reporting year, a manual process was necessary to map author affiliations to an institution from the CLARISA Control list. This process was both time-consuming and labor-intensive.

Now, with the new development implemented, these challenges are resolved. The PRMS Reporting Tool now predicts most equivalent partners for the author affiliations of any Knowledge Product at the time of their creation. This eliminates the need for users or the PRMS team (PPU, PCU, and D&D) to manually provide partner matches.

Details of the implementation:

When a Knowledge Product is created, most of the information is harvested from CGSpace through the M-QAP tool to reduce the amount of manual input required by users. Since CGSpace does not use the CLARISA Institutions list, the PRMS Reporting Tool saves the name of the author affiliations. However, with the new development in place, we now receive AI predictions from the M-QAP tool for all author affiliations. Each prediction includes a confidence level, which the PRMS Reporting Tool uses to determine whether the partner match for the author affiliation is automatically pre-filled or if it requires user input.

Matching types

Predicted by M-QAP AI ✨

If a prediction of a CLARISA institution has a confidence level equal to or higher than 90%, it is used to pre-fill the partner's section.

• If users consider the prediction to be incorrect, they can change the predicted partner to the correct one.
• Users must select the partner role for all predicted partners.

Example:

Manual match 🎯

If a prediction of a CLARISA institution has a confidence level lower than 90%, it is discarded and a manual match is requested to the user.

• These cases must be manually completed by users to achieve a complete status for the section.
• Only a few cases of manual matching are expected.
• If the user cannot find the corresponding partner in the CLARISA institutions list, they should submit a partner request. The institution will be included within three business days if accepted.

Example: