Page 49 - Crowdsourcing AI and Machine Learning solutions for SDGs - ITU AI/ML Challenges 2024 Report

Crowdsourcing AI and Machine Learning solutions for SDGs

• Temporal: Data may not be sensitive now but may become sensitive in the future due to
changes in context, such as shifts in policies and/or safety of specific populations.
• Relational: One dataset on its own may not be sensitive, however, it could become
sensitive if analyzed in combination with other datasets.

4 Classification

Based on the risk assessment, classification must be carried out at the dataset level to identify
which of the above data classification categories the data belongs to.

5 Standards, metadata, and documentation

For data sharing to be a success it is important that data are prepared in such a way that those
using the dataset have a clear understanding of what the data mean so that they can be used
appropriately. To enable this, data owners are encouraged to include with the dataset all the
necessary information (metadata) describing the data and their format. This information should
include such information as
• the methodology used to collect data
• definitions of variables
• units of measurement
• data format
• file type of the data
• any assumptions made

6 Data Sharing Guidelines

The figure below shows the steps to be considered when an entity (data owner) is planning to
share data for the ITU AI/ML Challenge.

41