Smart Matching Tool

Matching is considered as “one of the most important success indicators of an effective buddy project”. The smart matching tool makes a smart matching between mentors (members of the host community) and mentees (newcomers) using machine learning algorithms based on developed criteria.

Smart matching tool is a machine learning application.

The first step in developing the smart matching tool is to define the matching criteria to be used in social mentoring. According to HIVA, following criteria might be used;

- newcomer’s needs and goals,
- mentor’s offer and expectations of mentoring,
- mentor’s skills and professional background,
- mentor’s knowledge,
- interests and hobbies,
- language,
- availability and time commitment,
- geographical location,
- age,
- gender,
- marital status/family,
- personality.

Three different questionnaires were developed and embedded into the Smart Matching Tool in order to collect data in line with the matching criteria mentioned above. These questionnaires are applied to the mentor/mentee candidates by the coordinators during the intake. Criteria are also prioritized.

Three layers of matching are identified while developing the smart matching tool;

- AI layer,
- rule engine layer,
- manual matching layer.

Cosine similarity algorithm is being used with the data that has been collected from questions of the questionnaire.

Some other criteria feed the rule engine layer. For example, a woman mentor would like to be matched with a woman mentee. This request is satisfied with the rule engine.

Third layer is the manual matching. No matter what the smart matching tool proposes, the coordinator is always in the loop and has the final word (human in the loop concept). This is also useful in cases where limited data exists.

Smart Matching Tool processes personal data. Thus it complies with the GDPR.

Co funded by the EU

This project has received funding from the European Union’s Asylum Migration and Integration Fund (AMIF) under grant agreement No:957978.