Improving employment and work productivity has long been an ambition of the UK government, with the Department for Work and Pensions (DWP) handed the task to try and improve opportunities across society. In 1999 the government started investigating the benefits of joining up different online work and employment resources - the result was Worktrain.gov.uk, a portal providing access to 400,000 job vacancies.
By providing a single search facility for jobs, training courses, voluntary work and local childcare provision, the government hoped to create a resource to help people find jobs, re-skill and organise child care.
Real-time data from sites such as Jobcentre Plus and Learn Direct helped make Worktrain the UK's fifth most popular government website, but in 2003 the DWP, working with systems integrator Xansa, began a project to improve usability, introducing the first natural language processing system on a government site.
What were the business objectives of the project?
'The site brings together more than 6,000 job opportunities a day from Jobcentre Plus and a similar amount of training opportunities from Learn Direct,' says Graham Jack, project manager for Worktrain.
'Childcare and voluntary work opportunities are also available on the site, but we wanted to improve the usability to make it easier to search for opportunities.'
'Often you needed to know the terminology to be able to find the jobs, so we wanted to create a natural search engine and thesaurus that would allow our users to search in the language relevant to them.'
For example, he says some schools call dinner ladies 'midday meal assistants' and lollipop people as 'school crossing attendants'.
'The key objectives were to allow people to use their own words to search for jobs and also make the site easier for people to use.'
Jack says he also wanted to improve the way search results were presented, so they could be ranked by geographical or occupational preference.
'We had a wonderful resource but we needed to adapt it to make it easier to use and more intuitive.'
What were the key milestones in the implementation?
Jack says the first major milestone was gaining approval for the upgrade in mid-2003.
'Then we worked with Xansa and the University of Sheffield to help us develop the natural language interface,' he says.
The Institute of Employment, based at Warwick University, also worked with information handling specialists Adset to develop the thesaurus.
'It was something that we could have done in-house, but it would have taken much longer,' says Jack.
The prototype of the natural language search engine was launched on a trial basis in November 2003, becoming fully available in May 2004.
'We took it out to do some user research and the results were favourable,' says Jack.
'We then made some changes based on the results of user feedback.'
What technology was used?
Systems integrator Xansa worked with the in-house Worktrain team to implement the project, alongside the University of Sheffield, Adset and The Institute of Employment, to develop the natural language search engine and thesaurus.
A Microsoft .Net framework was used to develop the site, with eight SQL Servers clustered to form the back end of the system.
XML was used to take real-time job entries and integrate them into the search engine.
Worktrain and Xansa also worked with Jobcentre Plus's systems integrator EDS to ensure systems communicated with each other.
How did you manage the business change and people issues?
'The people it affected the most was our users,' says Jack.
'So by previewing the new functionality before the launch we were able to educate many of the users and ensure it worked.'
A regular series of meetings were held with technology partners to make sure they understood the implications of the upgrade and what the benefits were.
Jack says the key was to manage the project through a risk mitigation structure.
'Through mitigation plans we could ensure we delivered on schedule and to the quality we wanted.'
What results were achieved?
Following the upgrade, Worktrain.gov.uk has been visited by up to 20,000 unique users a day, making 450,000 individual daily search requests.
The natural language search engine and thesaurus has also improved the ease of locating jobs.
More than 6,500 official job terms have been added into the search engine, with numerous variants.
'We had to create 64,000 different variants and, because of people using plurals or mis-spelling, we had to create 856,000 different ways of expressing words,' says Christine Gould, senior solutions architect at Xansa.
What were the lessons learned?
'You can't underestimate the amount of testing you need to do to ensure the service is good,' says Jack.
When testing it is also important to use actual customers, rather than development staff.
'They can find and look at things in a different way to the development team,' he says.
Good communication with technology partners and content providers was also crucial, says Jack.
'One brick falling out of the wall could lead to the whole wall falling down.'
What were the business benefits and return on investment?
'The major benefit is that we have delivered a much better careers and learning service to UK citizens,' says Jack.
'It makes it clear that there is a lot of information available to them and it's easy to get to.'
Jack says the service has also helped eliminate barriers restricting employability.
'Finding child care or gaining experience can be a barrier, but these added services have helped with this.'
How do you plan to build the project further?
'We are looking at how we can make the service more readily available,' says Jack.
Electronic job kiosks have already been installed in London, through an initiative involving Worktrain and Jobcentre Plus, with more than one million job requests already made using this channel.
Other channels are being investigated including interactive TV and voice portals.
Computing says:
Searching thousands of web pages for a specific piece of information can be often be harder than finding a needle in a haystack.
So, by creating a natural language search engine that can help even the most non-IT-aware user find the career advice or childcare they need, Worktrain is helping to make egovernment services more easily available to UK citizens.
Expanding the service to other technology platforms such as kiosks and interactive TV can only help improve the service.
Project at a glance
- Worktrain.gov.uk, is a Department for Works and Pensions portal, providing information on 400,000 job vacancies, training courses, voluntary work and local childcare provision
- In 2003, Worktrain and systems integrator Xansa, started a project to introduce a natural language search engine to improve usability
- Worktrain.gov.uk is now visited by up to 20,000 unique users a day, who make 450,000 individual daily search requests
- More than 6,500 official job terms have been added into the search engine, with 64,000 different variants and 856,000 ways of expressing words
- Electronic job kiosks have now been installed in London, through an initiative involving Worktrain and Jobcentre Plus, with more than one million job requests already made using this channel
- Future plans to roll out to interactive TV and voice portals
Quelle: VNunet, 08.09.2004
