We were recently tasked with identifying Russian intelligence officers among the diplomatic staff of the Russian embassy in London using only open sources.
To help with this request, we used QIIK – the Quant Intelligence Information and Knowledge analytics prototype platform. The tool allowed us to rapidly build a consolidated repository of Russian diplomats based in London and Washington DC from openly available sources.
For this research, we had to look at dozens of sources, all with different formats and different type of information. Using QIIK data cleaning tools, we extracted and mapped the information we needed (job title, start date, end date). While we didn’t have the exact start and end date, we were nonetheless able to infer them using rule-based logic.
However, automation only took us so far. The human reasoning was still needed, even in these early stages of data gathering and processing. For instance, while the platform has functions to reconcile duplicate or near duplicate individuals, it did not pick up all the duplicates. Using the human meshing capabilities of QIIK we quickly reconciled the few remaining duplicates.
Once the initial data set of diplomats was ready we ran a couple of simple queries to identify the Russian diplomats likely to have been expelled by the UK government in March 2018 (the US data set was too unreliable to draw strong conclusion as we had some large data gaps in 2015/2016 (I.e. the lead up to expulsion ordered by Obama). Those queries returned seven names.
Building on this initial insight we ran the usual battery of searches to find out more about those individuals. While not all searches yielded useful results, we nonetheless found further evidence that some of the people identified were likely intelligence officer. One such example was Captain Igor E. A long-time staffer at the London embassy (since at least September 2014), Mr E. was scheduled to attend a commemoration ceremony on the 4th of May 2018 but didn’t show. As for all other data gathered for this investigation, this information was added to QIIK.
This case study showed the value of QIIK when investigating a large amount of data. Our data gathering and cleaning tools allowed us to rapidly build an institutional knowledge around a given issue. But this research also highlighted one of the fundamental problems with the automation of intelligence analysis: data are mostly unstructured. And while we have developed a tool that helps structure unstructured data, human insights and inputs are still required to leverage the full insights of data. Thankfully, with QIIK human meshing facilities, these human insights and inputs were quickly added and leveraged within our platform to produce the list of likely intelligence officers.