The GOV.UK product team in the Government Digital Service (GDS) has been testing a machine learning algorithm to generate links of related content during a search.
It says it should help users to find information that comes from sources beyond GOV.UK pages, otherwise known as its ‘Whitehall’ content.
Senior product manager Ganesh Senthi has outlined the project in a blogpost, which says it is part of the effort to help users move on from information they have landed on easily, largely from an external search engine.
The Whitehall content accounts for about 98% of what can be found on GOV.UK, as opposed to the ‘mainstream’ content that accounts for just 2% but 57% of the page views.
Senthis says the machine learning algorithm is aimed at identifying the Whitehall content and related links, giving content designers much more time to make sure it is well written and user focused.
GDS has A/B tested three algorithms across all of the GOV.UK content, looking for signs of them generating an increase in clicks on related links. This led it to identify one through which it has now run all of its content over three weeks to obtain a view of user journeys.
New releases
This was followed by releases of new versions of the algorithm, and the addition of over 400,000 pieces of content on GOV.UK.
The blog says the team is going to add further related links to Whitehall content and monitor performance.
“This is the first time that GDS has tested, built and released a machine learning pipeline, and there’s a lot of learning to be shared with the rest of the organisation and wider government,” Senthis says.
“The broader implications for helping to deliver smarter, more efficient public services are substantial.”
Image from GOV.UK, Open Government Licence v3.0
