DUME Project

Project DUME aims to change the centralised nature of digital platforms like Tidy City by:

1. Extending Solid Protocol that adeptly manages large-scale media datasets with decentralised web platforms.

2. Validating the robustness of the decentralised, user-centric features of the protocol created in 1., by implementing and testing it in Tidy City.

The challenges of implementing a protocol capable of handling vast volumes of high-resolution photographs with associated metadata for streets and roads, ready for AI model training, are manifold:

• For AI model training, swift and efficient data retrieval is essential. This involves implementing parallel data retrieval processes, batch data requests, and caching mechanisms to ensure minimal latency;
• The protocol should support rich metadata annotations, ensuring indexing mechanisms deliver quick searches and filtering based on this metadata;
• Specific AI access control ensures that AI model trainers can access the requisite data without compromising other personal or sensitive data. This involves creating specialised 'views' or 'profiles' of the data specifically curated for AI training;
• Given the dynamic nature of data (new images are added constantly), the protocol must support robust versioning mechanisms, allowing users and applications to fetch historical data versions efficiently.

DUME aims at creating a validated path not only for Tidy City's evolution but also for numerous other projects that depend on large amounts of media data, towards a decentralised digital paradigm where users maintain sovereignty over their media contributions.