The multi-agent ranking framework (MARK) aims to provide all the building blocks that are required to build large scale detection and ranking systems. Namely, the framework provides the following key components:

  • a distributed storage suited for very large datasets and BigData applications,
  • a web based visualization and management interface,
  • a distributed execution framework to reliably execute the detection algorithms and balance the compute load between multiple servers,
  • and an easy to configure triggering mechanism that allows to execute the detection algorithms when new data is available or when configured conditions are met.

All these components are implemented in a generic way. They are independent of the type of data that will be processed by the framework. This allows the data scientist to focus on his core business: developing effective detection algorithms.

Check GIT repository