How one council used machine learning to cut translation costs by 99.6%

Written by Sarah Peña, Head of Emerging Technology and Business Improvement, Swindon Borough Council

The United Kingdom (UK) has a strong track record of impactful innovation. Our public services need to find creative ways to deliver services differently and more efficiently so we meet our communities’ changing needs, while always protecting the people who rely on these services.

Swindon Borough Council serves a multicultural community of 230,000 citizens in the south of England. We continuously review emerging and evolving technologies, seeking to leverage them to reimagine and improve services, lower costs, and enhance efficiency. The success of a recent translation project that harnesses the power of artificial intelligence (AI) and cloud services is a prime example of this. My colleagues and I share our experience in this AWS Institute video.

Watch now: How Swindon Council used machine learning to cut translation costs by 99.6%

Swindon Council caters to a large number of languages spoken across the borough of Swindon. But this posed a particular challenge for the budget of our paediatric therapy team, where we were consistently overspending on translating important healthcare documents for families with young children, yet only covering the legal minimum that we were required to translate.

All that has changed dramatically since building Translate, a machine translation application. Our emerging technology team used Amazon Web Services (AWS) to create a solution that translates content more quickly, cheaply, and simply than the human translation services we had been using up until then, while keeping children’s data properly protected and maintaining translation quality. We began by using Amazon Translate, a neural machine translation service from AWS; then, we collaborated with AWS solutions architects to add services for user validation and personal identifiable information (PII) needs to build our solution, Translate.

Translate has provided multiple benefits for our team since we launched it. First, the speed. Before we started using Translate last year, it could take as long as 19 days to get a document translated. Now, it takes us under 15 minutes. There is no obvious trade-off in quality. In the 12 months since implementing the programme, we have not had a single request for clarification about the documents translated.

Second, the cost savings are equally impressive. In 2021, we translated about 400 documents, each one costing an average of almost £160. Our annual translation costs of about £64,000 ($81,000) have since shrunk to £27 ($34), just 7p (less than 10 cents) per machine-translated document. Plus, the council did not have to invest any capital upfront, beyond the time of our tech team. The return on investment for this project verges on the surreal: somewhere in the region of 6.3 million percent.

Third, the application’s simple, user-friendly design makes it accessible for all our staff. It takes about 15 minutes to learn how to use it. In less than a month, our entire paediatric therapy department was using the new programme with ease.

At the heart of Translate’s success is its use of cloud services, which are key to promoting innovation cheaply and securely. For our needs, a serverless solution is far more resilient and requires less maintenance. More importantly, it opens up new possibilities. The cloud provides a safe place to experiment and explore ideas. And when we start to see potentially valuable results, it also enables us to share our work with others in order to pool resources and build more quickly.

The positive effect of this project extends far beyond the time and money saved by Swindon Borough Council. We have open-sourced our Translate solution and it can be applied to organisations with similar challenges all over the world—not just other local authorities, but also hospitals, schools, central governments, charitable organisations, and private sector enterprises. Inspecting bodies in the UK, including the Office for Standards in Education, Children’s Services and Skills (Ofsted), have praised the solution, while those in our justice system have praised it for supporting inclusion and accessibility.

We have developed a more generic solution with built-in security and user management so that it can be safely adopted by our peers in the public sector. We have received widespread interest across the UK, including councils from Edinburgh to Newport, Southampton to West Berkshire already using the solution, and now also from two councils in France. The UK National Institute for the Blind is exploring how this could transform converting information into Braille.

We collaborated with City Trax Solutions, an AWS Partner specialising in empowering cities and local governments with cutting-edge, open-source technology, to the Translate solution. They managed deployment and testing, and delivered training and ongoing support. The solution also meets data compliance requirements. European cities such as Warsaw, Poland, and Rotterdam, the Netherlands, are now interested in this solution.

We’ve seen users apply the technology in innovative ways since it can be used across a range of council services. At Swindon Borough Council, the Warm Welcome Team used it to support refugees from Ukraine and Afghanistan. The Live Well Team uses it to promote cancer screening clinics and child vaccinations. The Adult Community Learning Team uses it as part of no-cost English courses. We’ve also provided schools with rapid turnaround information such as letters that must be translated for parents about school trips. We’re exploring how Translate can support the ways the council deals with incoming communications in any language by offering almost instant translations (such as phone calls, video calls, in person, email, social media, chat bot interactions, and more).

The overall impact of the project for the council has been profound. It has shown how we can take advantage of modern technology services to solve real business and community problems, at pace and in innovative ways; it also shows the power of collaboration with local communities and the private sector to harness new technology for positive societal outcomes. That has enabled us to become digital pioneers, and benefit other communities beyond our own. Above all, it has enabled us to free up financial resources and provide better, more efficient public services.


Originally posted here

 Read More AI & Data

Comments are closed.