CounteR’s Third Pilot Allowed End Users to Test the Platform’s Version 2.5

in News
Share this publication

Between 13 December 2023 and 5 January 2024, the project team successfully conducted the third pilot of the CounteR Platform. This activity, which took place in a virtual environment, offered an opportunity for end users to test version 2.5 of the CounteR Platform.

This third piloting period started with a kick-off meeting for end users, technical and ethics-focused partners to elaborate on progress and receive detailed answers to their questions. INS and AST presented the specific testing goals for the pilot. They provided a guided tour of the newly integrated platform features, the user manual, and the rest of the instruction materials developed to ensure LEAs and ISPs are able to test them independently and remotely. Beyond the usual communication channels and due to the entirely remote nature of this pilot, AST created a Slack channel and provided continuous assistance to partners during working hours.

Pilot Three introduced a more comprehensive methodology for end users to test the accuracy of the language models. It was agreed that this approach would be used as a standard for Pilot Four, as well as for the final evaluation.

The kick-off session included training to upskill LEAs’ knowledge on common error analysis strategies used by NLP specialists when developing and fine-tuning language models. The session consisted of an initial explanation by INS followed by a live hands-on coordinated exercise where all LEAs and ISPs practised successfully the steps with sample datasets.

During the remaining days, LEAs tested the platform in their own time and pace and provided feedback on its performance through questionnaires and one-on-one meetings with the pilot’s coordinator, INS.

Previous Post
In Spring 2024, Counter Platform’s Version 3.0 Will be Released
Next Post
CounteR and UCSC to Hold a Conference on Terrorism Risk Assessment Instruments
You may also be interested in these topics
Skip to content