A recurring challenge for military and humanitarian planners attempting to understand and map the human security environment is how to obtain real-time, location-specific and contextualised information. Moving closer towards understanding the human security landscape from the perspective of affected populations can help ensure that human security planning and provisions are increasingly tailored to that local context.
News data, in particular, is one way in which light can be shed on the possible subjective interpretations of human security issues from the perspective of the local populations themselves.
At Trilateral Research, we have been looking at how location-based news coverage at the sub-national level can tackle this challenge. The HAMOC application sits within our portfolio of human security work in which we explore and leverage the use of human security ‘big data’ and analytics to support decision-making and understanding around conflicts and crises. Using text classification algorithms that we are building specifically for the purpose of human security analysis and planning, we can explore trends in human security-related news coverage across time, by sub-national location, by news source and, most importantly, by human security thematic issue.
Consider the following. We know that modern slavery and human trafficking (MSHT) is enmeshed in the fabric of many conflicts and crises today, as both a driver and a symptom (read more in our book if you’re interested in the MSHT-conflict nexus). However, decision-makers need greater insight than this broad axiom. It isn’t enough just to know that MSHT is a significant and prevalent security threat in a particular country; we need to understand the differences in the prevalence of MSHT across different regions in a country, how subjective interpretations of this may differ within and between local populations, and how this has changed across time and space.
Take a look at the line graphs just below the map. The first two graphs show the news coverage connected with the Kidal region in north-eastern Mali.
We can clearly see a spike in news coverage that’s been classified as relating to both modernslavery and childsoldiers at the start of 2012, which coincides with the coup d’etat in Mali that year. News trends relating to modernslavery and childsoldiers appear to broadly correlate across the 2012-2020 timeframe, albeit with some discrepancies, such as the spike in late 2018 for modernslavery.
Overall, these two graphs show two insights that we’d expect to see:
- First, with the violence and conflict in Mali following the 2012 there was a clear increase in news coverage relating to modernslavery and connected with Kidal.
- Second, the news coverage of modernslavery and the news coverage of childsoldiers are closely intertwined over the subsequent years.
Considering the prevalence of the use of child soldiers in many of today’s conflict and crisis contexts, this isn’t surprising. Although of course, we can’t know from these graphs alone the extent to which these two topics were embedded in parallel within news content, or whether we are looking at two separate topics trends; the reality is likely to be a mix.
Now let’s look at Timbuktu, a large region covering much of northern Mali.
Looking at the same two human security thematic issues – modernslavery and childsoldiers – we see once again that there was a spike in news coverage connected with Timbuktu from 2012 onwards (although note the small spike relating to modernslavery in 2010). Again, both topics seem to follow similar trends. But even by taking a cursory glance at these two graphs, we can see that the correlation isn’t quite as close as it was for Kidal.
For instance, while coverage relating to modernslavery climbed consistently during 2018, this pattern isn’t replicated on the childsoldiers graph, which instead has two spikes and a period of flatness between them.
This illustrates that news coverage classified as relating to childsoldiers did not increase in tandem with news coverage classified as relating to the umbrella term of modernslavery. This type of observation suggests a need for further investigation from a human security perspective. It could be the case, for instance, that an increase in other forms of modernslavery, such as organ trafficking or sex trafficking, were underpinning the increased news coverage of modernslavery during that period.
To finish off, let’s now look at a single thematic issue, modernslavery, but by comparing Kidal to Timbuktu. Immediately we can see a number of interesting insights. While news coverage relating to modernslavery increased for both regions from 2012 onwards, the spike for Kidal was far more prominent and the coverage even fell away shortly afterwards. In comparison, the coverage for Timbuktu climbed more gradually from 2012 onwards but remained more consistently higher through 2020. And while the 2018-2020 period showed only two spikes for Kidal, there were multiple spikes for Timbuktu. From a human security analysis and planning perspective, considering that these two regions border each other those news trends discrepancies would suggest interesting points for further research and exploration.
In conversations with stakeholders and clients alike, we emphasise that exploiting data to understand (and ‘de-complexify’) the human security environment requires more than a single type of data and more than a single type of insight. It requires understanding connections, interdependencies, and other relations that allow us to journey along the data -> information -> knowledge -> insight continuum.
This is why the HAMOC application is founded on the basis of leveraging multi-format, multi-source human security data, grounded in human security domain expertise, developing and deploying the latest analytics and data science techniques, and providing clear and accessible visualisations for our users. Moreover, leveraging insights from multiple data sources (including news) can help to check our assumptions about the correlations between different human security issues while also using our existing domain knowledge to better interpret that data.
In future blog posts we will look at some of the other ways that we’re using human security data to build up the ‘big picture’ view of the human security environment.
For more information about our work on human security, please contact our team: