Page 435 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

2�3 Future Work

The project is ongoing at the modelling stage. It is focused on developing and validating
algorithms and training the data model. 4.4-Productivity

In the next stage, if provided with additional resources, pilot countries will be identified to
analyse the data in depth, validate the algorithms and consider other factors of delays not
accounted for through data analysis.

A business intelligence report will be designed to combine the data generated from the
process mining and AI model.

A dashboard/web app will be developed to present the results to a non-technical audience
(decision makers) to address specific backlogs and provide policy recommendations.

Additional resources are required to implement the planned future work, including the
development of a tool to automatise the end-to-end data extraction, data clean-up, data
modelling and AI integration to generate the business intelligence report, and present the
information in a clear and concise manner to a non-technical audience, particularly for the
selected pilot countries aiming to potentially scale to the 50 country operations.

The project may be implemented across the organisation to benefit the approximately 50
country operations where UNHCR conducts RSD under its mandate. The integration of
additional countries will enhance the capability of the tool to identify efficiencies and propose
new approaches across countries.
The research and model developed may also be adapted to RSD case management systems
used by States in their national asylum procedures. With a backlog of over 8-million applications
by mid-2024 and a global average processing time of approximately 400 days, this case study
has great potential to create a global impact.

3 Use Case Requirements

• REQ-01: Case management system/case processing data respectful of data privacy with
authentication, where RSD processing data is accurately recorded.
• REQ-02: Internal secure servers to store the anonymised and aggregated data extracted
from the case management system.
• REQ-03: Big data processing architecture.
• REQ-04: Data clean-up to ensure the accuracy and relevance of the data extracted.
• REQ-05: Training and finetuning the model for AI integration.
• REQ-06: Web-App displaying a dashboard including inference interpretation.

399