Skip to content

Using social media to manage large scale events

How software tools provided real-time information to police during the 2014 NATO summit in South Wales.

Social Media icons, copyright Jason Howie

Between July and September 2014 the International Technology Alliance (ITA) Controlled English technologies were used to monitor community reaction to the NATO Summit held in South Wales on Sept 2-5 as part of a study led by Cardiff University's Police Science Institute.

The study tested the software tools in interaction with human researchers in the field, to:

  1. conduct exploratory analyses of how the tools could enable users to get "eyes on" emerging events
  2. collect evidence on whether rapid analysis of social media traffic could usefully augment decision-making on the ground.

The results were briefed to the UK College of Policing in November 2014 as part of their national post-Summit debrief.

Martin Innes

A key finding from our work was understanding how social media analytics affords new ways to provide an ‘always on’ community impact assessment. The highly localised, positive and negative public sentiment recorded throughout both the build-up and during the event demonstrated a capability to understand public reaction through both space and time.

Professor Martin Innes, Director, Crime and Security Research Institute and Director, Universities' Police Science Institute

Live assessments

The study can act as an example of how police forces can use social media analytics to carry out ‘live’ community impact assessments during large scale events. Researchers were able to measure public reaction to events surrounding the summit, such as the announcement of local school closures and national media headlines, which reported on a 'ring of steel' around the summit.

Part of the work also involved analysing the tone of tweets base on location. NATO organised events in Cardiff Bay, which included a display of warships and a fly-past involving the RAF’s Red Arrows, generated a far more positive public mood compared with the disruption experienced in the centre of Cardiff and Newport.

Using geo-location analytics could allow police to monitor in real-time the mood of crowds in certain areas during large-scale events. The software and methods used for the study were able to detect specific incidents including transport disruption and gathering crowds as they were happening.

Institute for Technology Assessment Annual Fall Meeting, Peer Review Report

“We applaud this work. They should continue to use real scenario opportunities, like the NATO Summit, to explore Controlled Natural Language concepts and how they relate to Natural Language, as well as testing scalability.”

Informed decisions

Professor Alun Preece, from the School of Computer Science and Informatics, said: “As these incidents were being identified, we could direct our team of observers in Newport and Cardiff to obtain accurate information, such as the size of crowds, their mood and where they were going. We were able to capture information in a knowledge base that can be employed to answer questions, to help make decisions, and to manage an ongoing situation like the summit in real-time.”

The work will be part of the College of Policing's What Works Centre for Crime Reduction, which was set up to provide robust and comprehensive evidence to guide decision-making on public spending.