Social media analytics improves emergency response in successful trial by West Yorkshire police

Members of the public are typically the first to arrive at the scene of a crisis, and their initial assessment is crucial in managing the response of emergency services. Traditionally this involves a labour intensive process of information from 999 calls being added to the police system and reported to the relevant department while the caller is instructed on what to do next.

Advanced data analytics could provide a more efficient alternative. Their potential was recently put to the test by West Yorkshire Police, who teamed up with Sheffield Hallam University and SAS to develop a mobile app that harnesses communication with the public to manage crisis responses. The research was part of a European Commission grant-funded project dubbed ATHENA.

"We thought it was right for a law enforcement agency and potential end user to be driving it forward to make sure it had a real end-user focus all the way through and that's how we became involved," says Jessica Gibson, the project manager of West Yorkshire for Innovation at West Yorkshire Police.

"Throughout the project we've helped organise and gather feedback from potential end users on what would be the user requirements for any kind of system like this in an operational environment."

The consortium developed a prototype app and command and control platform that analyses on-the-scene insights to coordinate responses. Users of the ATHENA app can send text, audio, images, and video and request help at the touch of a button. It also creates a backup mesh network across app users if a phone signal goes down that can bounce messages between them until it finds an available network.

ATHENA trial

The three-year project drew to a close in September with a final trial of the solution at the West Yorkshire Police Training and Development Centre in Wakefield.

Volunteers used the app and social media to report relevant information to the police in a live simulation centred on five different scenarios involving around 100 different people. An ordinary day in the town environment was set up, albeit with some shady characters driving a van around the area and collecting suspicious packages from a bin while they went.

Protesters clashed in a public order incident and there was violence on a bus which left a police officer injured. A terrorist attack took place involving a chemical weapon, a firearms incident occurred and a vulnerable person was reported missing.

Information from the app and social media was fed into the command centre dashboard throughout the exercise and assessed to identify the dangers and coordinate a response.

Read next: How Avon and Somerset Constabulary is using Qlik Sense analytics platform to fight crime

A silver commander ran the control room while communication team members monitored the interactions from social media and app users on separate screens featuring map views of the area and a column displaying messages.

Direct messages could be sent to the source of any information, and additional social media reports was incorporated using a specific hashtag to avoid any legal and ethical issues around mining private data.

Clicking on a pin on the map could identify location-specific information, and danger or safety zones could be drawn around a designated area.

Users of the app received varying information depending on their access level, and the ATHENA Logic Cloud was used to analyse data and predict the locations of specific people on a heat map.

Future uses

"It was a great success," says Gibson. "People were submitting so much good quality intelligence that the scenarios ended up running a lot more quickly than we designed them to, because the commander was able to make those decisions so much more quickly than he normally would be able to if he was just relying on calls coming in and that information being relayed."

Any member of the public can download and use the app, but the receipt of false information remains an unavoidable threat that only human scrutiny is currently capable of managing.

"That's true of any method communication," says Gibson. "You have got to rely on the professionalism of our staff to spot what might look like an outlier message, something that doesn’t match up."

The system can add credibility levels to messages by analysing keywords, but it only learns within the cycle of a specified time-frame because the rules applied to one incident could be a danger in another.

The technical aspects of the software had already been successful evaluated, and feedback questionnaires issued both immediately following the exercise and then again one week later showed the users were also impressed.

Further funding and refinement will be required to make the product market-ready, but Gibson is confident it could benefit other blue light services, and even civic organisations such as local councils and mountain rescue units.

"We have areas in West Yorkshire that are high risk flood areas and it happens year in year out," she says.

"We could use this as a way of notifying people, letting them know which roads are closed, which is the safe way to go, which schools are going to be closed, that kind of thing.

"It started out at proposal stage as quite a high level terrorism-focused idea. The more the project developed, it became obvious that rather than just for the worst kind of crises, it would be usable and useful in low-level crises as well and in all kinds of situations."

Copyright © 2017 IDG Communications, Inc.

Bing’s AI chatbot came to work for me. I had to fire it.
Shop Tech Products at Amazon