JISC: Middleware for Distributed Cognition
Project Team:
Colin Tatham – technical lead
David Gilks – database programmer
Howard Noble – project manager
Jeff Kahn – VUE project manager
Katherine Ferguson – Cocoon developer
Matthew Dovey – UDDI and SRW consultant
Robert Gilks – Interface developer
Tom Coppeto – VUE developer
University of Oxford – Learning Technology Group
Mapping to ELFLearning Domain Services
Reporting
Resource list
Competency
Activity author
Course validation
Sequencing
Tracking
Assessment
ePortfolio
Marking
Quality assurance
Activity management
Grading
Learning flow
Curriculum
Personal development
Course management
Mapping to ELF
Common Services
Harvesting
Resolver
Logging
Workflow
Mapping
Messaging
Content management
Rating / Annotation
Chat
Person
Calendaring
Alert
AV conferencing
E-mail management
DRM
Identifier
Search
Presence
Authentication
Packaging
Terminology
Federated search
Role
Metadata management
Whiteboard
Context
Scheduling
Metadata service registry
Filing
Service registry
Rules
Authorisation
Archiving
User preferences
Group
Member
Format conversion
Forum
Mapping to domain: repositories
• Distributed/ federated search– Ability to cross search multiple repositories with Z39.50 and SRW
protocols
• Discover– Ability to find the appropriate copy of a resource through OpenURL
mechanism
• Resource list– Ability to store resource metadata (reference) compliant with the IMS RLI
data specification
System interoperability: MDC project in context of repositories domain
Search(MDC Interface ) Distributed
search(JAFER )
Data layer - repositories
Middleware Layer
Learner InterfaceLayer
JSP Z39.50SRW
Dublin CoreLOMOAI
IMS RLI
UML activity diagram
MDC interface: http://jafer2.oucs.ox.ac.uk:8080/MDC/search.jsp
Resource list
Cocoon transforms - PDF
Cocoon transforms - XHTML
Cocoon transforms - XML
XBib: http://sourceforge.net/projects/xbiblio/
XBib http://sourceforge.net/projects/xbiblio/
Mind mapping with VUE
Sharing resource lists – store in repositoriestowards the semantic web
• Repositories holding resource lists can be targets for distributed searches
• Repositories storing mind maps can be interrogated to give learners indication of how articles are related
UDDI
Discussion
Framework for optimising recall and precision.
Cooperative Association for Internet Data Analysis (CAIDA): http://www.caida.org/home/index.xmlTaken from CAIDA Walrus project: http://www.caida.org/tools/visualization/walrus/
The information landscape
• Search and discover– Open searching vs. finding the appropriate copy based on learner’s subscription profile– Search metadata globally– Discover appropriate services that pertain to the required resource (full text, ability to edit etc)– Distributed search of OAI repositories– Optimising recall and precision:
• Recall illustrates the confidence that a search returns all the information you are interested in • Precision is the confidence that the results of your search are relevant
• Centralised vs. distributed repositories– Bandwidth usage– Searching across heterogeneous metadata– Functions above metadata e.g. annotation, ratings, related resources, – Ownership and authentication
• Additional components:– Caching– De-duplication– Matching algorithms based on search criteria– Parsing to improve metadata presentation
• Social solutions– Interconnected resource lists/ mind maps– Building recommendation into the core of search practice
• Usability– Use language of the learner not database or cataloguing– To search as part of a network of learners not alone
Links:
• Final report (draft): http://users.ox.ac.uk/~howardn/Publish/
• Interface: http://jafer2.oucs.ox.ac.uk:8080/MDC/search.jsp
