Jan Swillens: On the frontline in the greyzone and in war: From secrecy to action.
Abstract:
Facing shifts in the global balance of power, digital transformation, disruptive technologies and developing conflicts, the MIVD is on the front line both in the grey zone and in war. In his keynote speech Major General Jan Swillens talks about three crucial conditions to be able to deliver intelligence and security to our military, our society and our allies and protect our democratic values.
Mark Galeotti: Lessons in War: what have we learned from and about Russian intelligence since February 2022
Abstract:
(Not available)
Patricia Damen: Towards a data-driven intelligence service
Abstract:
The amount of available data for intelligence purposes is growing rapidly. Data can be considered as the oxygen for our intelligence process. No data, no intel.
To get the most out of our data, MIVD (and AIVD) are currently transforming to become data-driven intelligence services.
However, during this transformation process, technological developments in the outside world like Artificial Intelligence are also evolving rapidly. This can be seen as a threat, as for some countries these technologies are at the core of their efforts to gain Information advantage. On the other hand, these developments can be a positive change if we manage to incorporate them in our intelligence process in a compliant way.
In the Keynote, The Director of Data and Information Management of MIVD will discuss this journey with you, both from a technical perspective, as well as from a legal perspective. She will illustrate her presentation with examples to give you an insight into the challenges we are facing.
Ryan Shaffer: Unsupervised Machine Learning on Colonial Kenyan Intelligence Reports: History, Methods, and Challenges
Joint work of Dr. Ryan Shaffer and Dr. Benjamin Shearn.
Abstract:
This paper uses natural language processing (NLP) to analyse intelligence reports declassified by the Kenyan government. In doing so, this work argues for more robust incorporation of computational methods in the intelligence studies field to better analyse declassified records and understand how NLP can shed light on official records in new ways. It provides a historical case study to demonstrate how unsupervised machine learning can be leveraged to analyse security threats in a large body of intelligence records. Focusing on Kenya prior to the 1952 Mau Mau uprising and the ensuing State of Emergency, the paper offers a quantitative methodology to understand threats to law and order using a computational methodology to process over 13,000 pages of text. Though it acknowledges that colonial records have been lost or destroyed, the corpus consists of the most robust collection of declassified and digitized colonial intelligence reports from colonial Kenya. The paper uses named entity recognition (NER) to extract location names from the text and identify specific areas of concern in the records. Next, it utilizes a zero- shot learning technique to detect threats even if specific words do not appear on a page. Then it employs trend analysis to track threat issues over time. Finally, it uses these methods to locate events on a map with minimal human intervention. While demonstrating the usefulness of these methods for threat detection, the paper also discusses shortcomings and challenges with these methods, especially as they relate to history.