Share

The Organisation for Economic Co-operation and Development (OECD) realised two years ago that the way it was storing and delivering content was out of sync with the needs of its users.

The OECD is a think tank for social and economic policy which publishes roughly 400 books a year in many languages. This involves “a lot of data and literature, working papers, policy briefs”, says Toby Green, head of publishing at the OECD, with content going out to 15,000 subscribers, including universities, governments and businesses. (See also: what is a graph database?)

“Four hundred books a year doesn’t sound like much, but if you consider that we break them up into independent parts we are publishing 35,000 items a year,” he told ComputerworldUK.

“So we need an engine that can cope and sort through that so the user doesn’t get lost.”

Hub and spoke model

Green and the OECD wanted to move to a centralised content library with a "hub-and-spoke" dissemination model.

After assessing its options the OECD decided to move its data to a NoSQL database solution from MarkLogic.

The NoSQL database allows the OECD to store all of its content - from journals and books to charts and graphs - in one, standardised way. The built-in search capability allows users to query data and metadata for faster, more accurate content discover.

Previously the OECD had been using a home-grown relational database based on SQL, and distribution was handled by a team of content editors.

Half of the editors would be tasked with manually loading summaries into the publication database and then someone else was responsible for dissemination across digital channels.

Now they can upload summaries in 30 different languages in just three clicks. “The content manager is able to intervene and create everything and disseminate it themselves. They are getting full responsibility of the value chain and see the impact, and they like that,” says Green.

The distribution ‘spokes’ include home-grown channels like the OECD iLibrary, online bookshop, data.oecd.org and OECD.org; as well as third-parties like Google Play, Scribd and RePEc and specialist search engines like Summon and Ex Libris.

MarkLogic

The first project the OECD took on with the help of MarkLogic was to integrate its existing library of 4,700 book summaries in 30 languages into the new “hub and spoke” library platform, which it managed in under six months.

This proof of concept allowed the OECD to eventually move all of its publishing activity to the MarkLogic system. Green says the next stage in OECD’s content strategy with MarkLogic will be building out semantic tools for better meta-discovery.

Read next: JPMorgan consolidates derivative trade systems with MarkLogic NoSQL database

Couldn’t the OECD buy this sort of software off the shelf though? “How on earth do you manage all this disparate stuff?” asked Green.

“You can go to the market and buy a book management engine or a journal management engine but I don’t know any that support all of these data sets alongside each other.

“There is no engine that supports all four content types in one place. To build a unified user experience, with single search and seamless movement we needed to manage all of it in one place.”

The model clearly serves a niche need, as the United Nations (UN) has been working with the OECD to implement a similar content management system since last year. The deal with the UN was signed in October, built in beta in three months and in production within six months.

Read next: BBC iPlayer sees performance uplift after relaunching on NoSQL database

Alternative Routes

Why not go open source though and save on the cost? With just four full time engineers the OECD needed a solution that was low-maintenance.

“We wanted a solution that was maintained and robust in the long run," explains Pascale Cissokho Mutter, head of project and information management at OECD. "The problem is it’s not an open source culture here and the challenge is you end up having to do the engineering yourself and that becomes onerous in the long run.”

Read next: 11 reasons why open source is taking over enterprise

How about the big vendors? “We like working with companies who have momentum behind them,” Mutter says.

“We also tend to go with specialists rather than the big companies as we find you get a better, closer relationship and a better hearing. The big companies are impersonal and we aren’t big enough to matter to them.”

Find your next job with computerworld UK jobs