A complex data mapping application
Expertise
Design and Interaction
Platforms
Web
Deliverables
UI, UX
The Brief
The goal of the T3 Data Mapping project was to provide an interface that allows a group of data standards and schema organisations to work together to map their data standards to supporting Learning & Employment Records (LERs), a project within the T3 Innovation Network.
In short, we needed to build a universal translator to allow multiple educational data standards organisations with different tagging standards to speak to each other.
To make that possible a master schema was selected, the Common Education Data Standards (CEDS), to act as a base reference, internally called “the spine”. Representatives from each organisation provided experts and used the platform to map to the spine. Where no properties existed in the initial CEDS schema, organisations could add new properties to the spine. As new properties are added the new “synthetic spine” will grow over time.
Our task was to develop a platform to make the uploading of schemas as simple as possible, allow sorting, aligning and annotation of data standard schema terms and support multiple output formats.
The Process
Together with the client we implemented our Accelerated Learning and Testing Methodology to understand the underlying needs of the application, quickly integrate all of the previous learning by the expert team, diagram the full information and user flow, reverse engineer requirements and build testable prototypes for external testing.
This is our usual method for dealing with complex requirements and interfaces on a compressed timeline and budget. We focus on learning as much as possible as quickly as possible through accelerated iterations of the requirements, focusing on examining the data structure and extracting user requirements.
By running a string of short but intense working sessions we can quickly turn around diagrams, sketches and wireframes that demonstrate and test our understanding of complex informational relationships. In this case we were able to rely on a spreadsheet prototype that captured the complex data structures that we would need to allow users to manipulate through the interface.
After testing with internal experts we quickly moved to multiple sets of 1:1 testing interviews using static prototypes, where external experts were able to give feedback on the legibility of the process and clarity of the screens. During the time between interview sets we updated the prototype until we were confident that no more could be learned from the static prototypes and development on the alpha platform began.
As of September 2022 the T3 Mapping Tool is in the final stages of alpha development, and being tested in a small private pilot. The next step will be to analyse the user interaction feedback data and improve the product interface before a larger beta release.
Expert Tool with Incremental Gains
There was no doubt from the start that this was going to be a tool for experts. The description language that organisations use for their schemas requires an expert level of understanding of both the terminology and the concepts.
What was needed was a platform that could make the task a) possible at scale, but also to find acceleration wherever possible to make the data entry and manipulation easier and faster.
Our goal was to find efficiency in the process wherever possible, and as we worked through the data model we could see opportunities both in guiding experts through steps from large scale categorisation of data to fine grain mapping. We also discovered opportunities for using business rules and algorithms at the fine grain mapping level to anticipate certain tagging and alignment tasks to save time.
Chunking Tasks and Collaborating
One of the key discoveries was that most organisations will be collaborating to create their mappings. Multiple people will be involved in setting up projects, uploading data and doing the fine-grained alignment. For this reason we chose to break down the tasks into “chunks” of data, from organisation set up, to rough data uploads, coarse grain domain-level mapping and then fine grain alignment.
Because multiple experts will be working together the interface also allows for commenting on changes, maintaining a change log and notifications of status changes that can be shared by the whole team of experts in an organisation.
Incremental Gains through Algorithms
In this particular example the user is mapping a new component to the spine by merging multiple components into a single element. The system can recognize this as an aggregated relationship between a single element on the spine and the multiple mapped components. The relationship is automatically tagged appropriately, which saves the user from having to tag elements individually.
By layering up the different algorithmic cases business logic we are able to save time fractionally and increase the speed of the mapping overall.
Jeanne Kitchens
Associate Director
Center for Workforce Development