Collaboration, Coherence & Capacity-Building
Using DSpace to Support a Major Social Science Research Programme in the UK
Patrick Carmichael and Richard Procter
CARET, University of Cambridge
The Teaching and Learning Research Programme
• An eight-year, £30 million programme of Education Research• Administered by the Economic and Social Research Council (ESRC) • Charged with supporting and developing education research in the United
Kingdom• Over 40 projects • Over 300 researchers based in Further and Higher Education Institutions
across England, Wales, Scotland and Northern Ireland.• Research Fellowships• Research Capacity Building Activities• Thematic groups and seminar series• Outputs include books, special editions of journals, articles, research
briefing and reports, teaching materials and videos
Elements of the TLRP IT Infrastructure
• Programme website– Programme Information– Resources for participants and project administrators– News and Events– Project Summaries
• Project websites– Most run by participating institutions
– TLRP hosts some project websites
• Database of individuals, projects and organisations• Collaboration Environments: dotLRN and Sakai• DSpace
Characteristics of the TLRP DSpace
• Single Collection including– Conference Papers (49%)– Occasional Papers, Research Briefings and Reports (22%)– Books (9%)– Published Articles (20%)
• Ready for the RAE!• Sector descriptors based on British Education Thesaurus• Descriptive vocabulary developed by TLRP community• Submission at project or individual level• Moderation by members of TLRP Directors’ Team• Open Archive Initiative OAI-PMH Interface
DSpace content as RSS in the Sakai Environment
OAI feed is transformed into RSS feed which then is aggregated in Sakai
Using DSpace to Explore Collaborative Practices in TLRP Projects
• DSpace provides (through the OAI interface) a means of observing over time how ‘co-authorship’ networks develop within and between the projects of the TLRP.
• Co-authorship is only a proxy for collaboration, which may take different forms in different settings - and it may not be the best one.
• Large scale studies have been undertaken:
– Science (Newman, 2001-4)
– Sociology (Moody, 2004)
• These have established some key approaches and metrics which can be applied.
• But these have been ‘historical’ analyses, not ‘live’!
Bipartite Author-Publication Graph
Vertices represent authors and publications and arrows represent the relationship 'is an author of'.
Note how author A1, as well as 'single-authoring' publication P1, is also a co-author of publication P2 with A2 and A3.
Author Graph
Vertices represent authors and arrows represent the relationship 'has co-authored with'. The lack of this relationship between author A1 and
A4 is evident, as is the isolation of author A5.
With publications removed, the number of times any author has co-authored with another is not represented.
Fundamental Statistics for the TLRP DSpace
Measure Sept 04 Jan 05 Apr 05 Jun 05
Numbers of Authors 35 149 244 267
Numbers of Papers 33 97 178 227
Mean papers per author 3.94 2.33 2.64 2.86
Mean authors per paper 4.18 3.58 3.62 3.36
Collaborators per author 5.88 4.88 8.19 7.88
Collaborators per author (adjusted)
n/a n/a 5.03 5.05
A component of the author network
Christie, D., Tolmie, A., Howe, C., Topping, K., Thurston, A., Jessiman, E., Livingston, K., and Donaldson, C. (2004) The impact of collaborative group work in primary classrooms and the effects of class composition in urban and rural schools (TLRP Annual Conference, November 2004)
Howe, C.,Nunes, T.,Bryant, P., and Jafri, S (2004) 'Intensive quantities: why they matter to mathematics education' (TLRP
Annual Conference, November 2004)
A portion of the author network, showing cohesive subgroups, cliques and isolates
The author network a month later showing the development of a ‘giant component’
Conclusions and Next Steps
• OAI can be used in many different ways!
• But care needs to be taken when using something designed for resource discovery for resource description and in more complex applications.
• Careful design of metadata frameworks and vocabularies and aggregation of metadata records allows exploration of complex phenomena.
• Caution!
– Small dataset as yet
– Cannot be sure we have ‘complete’ set’
– Might need to look at other bipartite graphs, or at combinations of several.
• Just because you can, doesn’t mean you should!
Thank you!
• Contact Addresses:– [email protected]– [email protected]
• Websites:– http://www.tlrp.org– http://www.tlrp.org/dspace