Hackathon for the Ministry of Defence

What if UK’s Ministry of Defence could harness the power of AI and Machine Learning?

author photo
Written by Objectivity
Innovation in the military

Harnessing Machine Learning for the MoD

Ministry of Defence needs to efficiently process information to be effective – that’s a given. But how can you organise clearance rights and disseminate in time key information across multiple countries and sectors? We have tackled this problem during a hackathon in London, 29th-30th November.

Currently, MoD’s Defence Logistics department collects huge amounts of data from various sources (sensors, operational and engineering activities, logistic, financial and supply chain transactions). This data has to be shared with an army of people (pun intended), also from various countries, who have varying access rights. Doing this manually? Very time-consuming. A more detailed description of the problem is here.

What if AI and Machine Learning could apply business rules and instantly share important information accross many departments? More importantly – how such a solution would impact the business world?

The goal was: given that you have a dozen of databases to work with and a list of thousand users (each with specified name, organisation, role, clearance level, etc.) – how do you apply business rules so that people on various roles can quickly access information crucial for their work, without compromising confidentiality of data?

This was no small feat, given the extent of dependencies, but we were happy to tackle this problem and came up with an interesting solution.

We used the Prolog language to implement the business rules important in this hackathon. Then we analysed available databases to apply the business rules. We performed the analysis from both sides: scanning the database of 1000 users and their credentials, and scanning the databases with information to be accessed. We searched for key words and data structures that helped us to define a confidentiality level of the given entry. This allowed us to create a matrix of dependencies and clearances for each of the 7 roles defined to be using the theoretical solution. Then we created a UI visualisation for 2 of these roles and shed some light on how the full solution could look like. One of the enhancements we proposed is an engine for implementing new business rules: it would interpret rules written in natural language and implement them in the solution’s logic.

This was an incredible experience and we are grateful that we could participate in this event. We have learned that one of the current challenges for the public sector is large amounts of unconsolidated data, that need to be shared and processed across many departments with varying clearance rights. This problem, also prevalent in the business world where a lot of separate systems work inside a single company, could be solved using AI and Machine Learning, which we have proven.