LAMP (Library Analytics and Metrics project) is a partnership between Jisc, Mimas (at the University of Manchester) and the University of Huddersfield. The aim of the project is to spec’ and develop a prototype shared library analytics service for UK academic libraries.
It’s carrying on the work of previous library activity data projects, including several to which the University of Lincoln contributed data:
- LIDP (Library Impact Data Project)
- SALT2 / the Copac Activity Data Project
- RISE project – Lincoln wasn’t involved in this one!
- Activity data synthesis project – see also: http://www.activitydata.org/
Lincoln intends to contribute some of our data to the LAMP project, too.
At Monday’s meeting, David Kay of SERO Consulting spoke about some work being done to categorise the use cases for the use of analytics data in HE, and identified tools using library analytics data in stages from proposals [many] through pilots [some] to production services [few. In fact, probably one].
He also discussed the legal / ethical position in using student activity data to build [shared] services, and how we need to move beyond a position of fear of making use of this data in case “someone” objects. In fact, as Richard Nurse (Open University) highlighted, the attitude from universities (edit: and the attitude of students!) is more likely to be:
“You’ve had all of this data for ages, so why haven’t you used it to improve the student experience?”
A number of people from MIMAS spoke about some of the technical and data challenges. Some useful recent blog posts:
- Planning the LAMP Architecture (API-driven development, and incorporation of external data)
- Unique Identifiers Which Don’t Identify Anyone! (hashing IDs)
There are a lot of issues around data normalisation (combining and comparing data from multiple HE institutions), and around the problems of dividing user data into a large enough number of categories to allow for meaningful use of the data, but few enough categories to permit statistical analysis.
Some potential problems for Lincoln are going to be:
- How do we convince sceptical parts of the University that this is a legitimate use of student data? What are the ‘norms’ for anonymisation and sharing? Is there a role for some targeted advice from Jisc Legal?
- Related: how much should we be normalising our own data before we release/publish/share it? It needs to be genericised (is that a word?) enough to reassure the University and individuals, but detailed and ‘real’ enough to allow normalisation to be done centrally, and to permit useful comparative uses of the data across institutions.
- Is there a standard for generating anonymised unique IDs? If so, where’s it documented? How do we make sure that the same processes for extracting and processing data are used year in, year out, so that individual (anonymised) users can be identified, and comparisons made, across multiple years?
- We’re currently suffering a (temporary, I hope) developer ‘dry patch’ at Lincoln. How do we make sure we get and keep the technical skills to maintain this data and the extraction/manipulation/publishing processes?
Some of my concerns I fed into a useful and interesting mock ‘post-mortem’ exercise after lunch.
I’ll let the LAMP blog publish the detail of the exercise, but in summary we all imagined ourselves in the year 2015, looking back on a LAMP project which had failed catastrophically. Then we tried to capture as many reasons as possible why it went (hypothetically!) wrong.
Everything from data quality/UI problems to legal challenges against our use of the data, and alien invasions disrupting Jisc’s plans… ideas all went onto Post-it notes and were sorted into the following broad categories on a wall (see photo):
- Data quality
- Costs & processes
- Doomsday (the aforementioned alien invasion)
These presumably will feed into the project management, to try and design out or mitigate against as many problems as possible, and have plans for dealing with them if they do occur.