Shared Data

Library systems create vast amounts of data. Every time a web page is viewed, a book is loaned, a user authenticates, a fine is paid, or a journal is accessed, another piece of information is written to a database somewhere. When the issues surrounding Data Protection came to the fore, systems administrators spent huge amounts of time and effort in locking data away from prying eyes. In some cases to such an extent that the data became unusable and irrelevant. Since then we have realised the potential of gathering statistics from this data. Loan stats show us which books we can withdraw, authentication stats show which services are being used, and journal access stats help us choose which packages to purchase. However there is a growing trend to utilise this data beyond the confines of the institution.

Last night I had one of those 'big brother is watching' moments when I realised that Kent County Council was following me on Twitter. However on closer inspection it was a new division of the Council called 'Pic and Mix'. This new service is devoted to making Council data freely available to anyone that wants to utilise it in some way. The Pic and Mix web site offers statistical data and informational data in areas such as waste, heath, business, and even libraries.

University Libraries are also following this trend in making their data available. The MOSAIC project was developed from the recommendations of the JISC TILE project. In a nutshell the MOSAIC project's aim is to develop a recommendation system based on the collected usage statistics of a number of institutions. The project even had a competition running with a prize to the individual or team that produced the most innovative application for this shared data.

JISC is also involved with the COUNTER project. This project is a collection of individuals and organisations that believe the collection and usage of online resource statistical data should be performed to an agreed set of standards. In this way any Library, Vendor, or Institution could produce or utilise data from anywhere else in a standardised way. This project will hopefully pave the way to a more open and simplified approach to a data sharing. JISC has also just announced the start of a new project that is an extension of the COUNTER project. This project is concerned with extracting statistics at article level rather than just journal level...see here for details.

So what is all this shared data being used for?

A new book (Library Mashups) by Nicole Enagrd explores the way in which data is being used by libraries. Nicole and other authors explain how libraries are combining freely avaible data from many different sources to create mashups of data for staff and student consumption.





A really good examples of shared data usage by a commercial company is bX from ExLibris. bX takes usage statistics from customer SFX installations and uses this data to compile lists of recommendations based on search results. This means that links that appear within the SFX menu will not only be recommendations based on context, but also based on relevance.




There are also a whole host of other home-grown examples of how shared data is being used in Universities, many of which have come off the back-end of projects like MOSAIC. There are even more examples of how recommender services can be built in-house using just the institutions own data. The concept is very simple:

1. The user clicks on an item in the OPAC


2. A lookup is on the data finds anyone else who has borrowed that item in the past


3. A second lookup compares the items borrowed by these users


4. A list of the top 5 items borrowed by these users is displayed
   

In this way we can recommend the top 5 other titles that have been borrowed by users who have also borrowed the original book the user searched for. This is very similar to the way in which Amazon displays it's "Customers who bought this item also bought" feature. A good example of this working can be seen on the Huddersfield Uni OPAC, under the 'Print Copy/Holding Information' box, a list of recommended titles is displayed.

As the methods and uses of data become more standardised and institutions realise the potential of shared data, so more services will become available. Allowing our data to be shared will not only benefit others, but will directly benefit us as new services are developed to make use of the data.