Capita build Linked Data Libraries
Capita’s market-leading online library catalogue product, Prism, is built using Linked Data and allows others to build applications on top of it. We spoke to Head of Development, Andy Latham and Technical Lead, Phil John to find out how and why.
Prism is a critical product for Capita’s Library Business. Libraries running this product serve over three million borrowers in public and university libraries up and down the country and it is a vital part of their customers’ services. The latest version of Prism comes from a strong pedigree dating back to the first web-based catalogue product in 1994. The latest version, Prism, is delivered as software-as-a-service (SaaS) and pushes the product into the next generation of web technologies by employing Linked Data approaches.
The Talis Platform meets the cost, quality and scalability requirements of our powerful SaaS discovery interface
Using Linked Data inside the application allows Prism to store, query and represent bibliographic data in a far more sophisticated way than the long-lived and widely used MARC standard. This sophistication allows Prism to provide a service that supports visitors in choosing their own navigation through the catalogue, promoting serendipitous discovery of what the library has to offer.
Linked Data is able to do this by replacing the record-centric view of the world with a web of data that recognises different aspects, author, title, publisher and so on, as equally important things that users may search for.
Over three million people generate a lot of traffic for such a critical application as this. Rather than scale out database servers themselves the team based Prism on top of the Talis Platform and rely on hosting provided by Talis Consulting. This combination allows the team to focus their efforts where it matters most; the data and Prism’s presentation. Prism hosts configuration data for its individual customers in the platform as well as large amounts of bibliographic data, making it easy for the Prism application to configure itself for requests from different customers.
The scale of this Linked Data application acts as an excellent proving point for this technology, with Prism receiving more than 1.7 million hits per day on relatively modest hardware. Those requests rely on the underlying data hosted with Talis and generate over 50 million transactions per month to the Talis Platform. Andy Latham, Head of Development for Capita Software Services, says “The Talis Platform meets the cost, quality and scalability requirements of our powerful SaaS discovery interface.”
The Talis Platform abstracts away all of the complexities involved in enterprise searching allowing us to focus on evolving the powerful user interface
To provide an even faster experience for users of Prism, the application uses several caching strategies. Caching within the Prism application itself provides a first level of caching for commonly requested pages and assets such as images and CSS files. The second level is at the user’s browser, where Prism provides the browser the information it needs to make optimal caching decisions for the Prism pages.
Prism uses caching to improve performance by reducing the number of requests it makes to the Talis Platform. This isn’t always possible, especially for searches where requests are often very specific to an individual’s search. As search is the key feature of Prism it is important that even these un-cached requests are served swiftly.
Prism also relies on the Talis Platform for its search capability, with the underlying platform generating comprehensive indexes from the catalogue and providing a rich search capability. “The Talis Platform abstracts away all of the complexities involved in enterprise searching allowing us to focus on evolving the powerful user interface” says Phil John, Technical Lead on Prism. By relying on Talis to provide scale for the search features, the Prism team saved time, effort and cost they would otherwise have had to spend developing and testing their own scalable search services.
Search results also have to change as stock is changed, with new stock arriving and old stock being retired. The Prism team solved this by building a simple RESTful API that takes changes from the library’s local catalogues and mirrors those changes into the catalogue stores. Libraries using Prism make over 9000 changes every day and these are reflected in search results within minutes. With some catalogues containing over 2.1 million bibliographic records that’s performance that keeps Prism’s users, and customers, happy.
Print this page
