Itembanking Infrastructure: A Proposal for a Decoupled
Architecture
Mhairi McAlpine and Linn van der ZandenScottish Qualifications Authority
CAA Conference
Introduction and Definition
This presentation outlines a decoupled architecture for an itembanking system. Using small interlinked pieces of software, coupled with strong workflow and access management, it describes how an itembanking system may function as a decoupled system.
An itembank is considered to be made up of a repository and database. It is a piece of software which contains and holds details about items.
An itembanking system is considered everything up to the point of delivery, and also everything from the output of data from the delivery system. This is comprised of software; workflow processing and access management.
Decoupled Architecture: Benefits
Easily adapted to accommodate change in model or workflow
Pieces can be upgraded in line with specifications
Small chunks can be built and used immediately
Sophistication of the system/ community will place additional demands
Decoupled Architecture: Implications
Strong standards compliance required
Additional functionality to support existing processes may not be required at a later stage
Robust testing required at element level
What does Itembanking entail?
Software components Storage of items
Generation of items
Delivery, Marking and Result Processing
Test Construction
Data and user management Workflow processing
Role-based user permissions
Storage of Items
The Itembank:
Repository storage of QTI files, resources, manifest files
Database storage of LOM and QTI metadata, search and retrieval functionality
Unpackaging Functionality splitting incoming content packages into their respective parts
Generation of Items
QTI Authoring Tool: outputs QTIv2.0 items
Specialist Authoring Software:to create additional embedded elements (eg graphics)
Metadata Tagger: attributes data to items according to IEEE LOM standard and extracts QTI metadata
Application Profile Development Software: interface for development of profiles which prefills metadata
Content Packager: packages elements of QTI items according to IMS specification
Test Construction
Glossary Development Software: produces glossary which defines statistics for test construction
Test Construction Software: takes in application profile (metadata) and glossary (statistics) to produce algorithm
Item Analysis Software: runs required analysis from algorithm, identifying items from pool that match the conditions
Delivery, Marking and Result Processing
Delivery Software: imports QTIv2.1 package from itembank, and sends responses to Marking Processing Software
Marking Processing Software: has several elements for different item types i.e. human based, computer based or both
Result Processing Service: aggregates items and implements pass mark or grade boundaries
Master Results Databank: stores candidate interactions with item IDs which are fed out from the Delivery Software
Future areas of Exploration
First attempt at scoping the potential for decoupled architecture
Requirements must be scoped for each element
Roles of users need further defined
Workflow processes need defined
We are currently exploring collaborative development of an itembanking infrastructure with a number of organisations, both in Scotland and further afield, recognising the generic nature of the development required and the benefits of ensuring high quality data transfer between systems.
Contacts
Mhairi McAlpine
Project Manager
Linn van der Zanden
Learning Technologist