Upload
fridolinwild
View
11
Download
1
Embed Size (px)
DESCRIPTION
Citation preview
Bringing together what belongs
togetherFridolin Wild1), Xavier Ochoa2), Nina Heinze3), Raquel Crespo4),
Kevin Quick1)
1) The Open University, UK, 2) ESPOL, Ecuador3) KMRC, Germany, 4) UC3M, Spain
Outline
• The Idea: – Spot unwanted
fragmentation– recommend a
flashmeeting• The Data: ECTEL,
flashmeeting• The Method(s)• First results• Evaluation
Before we begin…
Information could be the quality of a certain signal.Science could be about systematically giving birth to information in order to create knowledge
Information could be a logical abstractor.Knowledge could be the delta at the receiver (a paper, a human, a library).
Science, Information, Knowledge
(96dpi)
Science is made in networks
Researchers (people, artefacts, and tools) in various locations with heterogeneous affiliations, purposes, styles, objectives, etc.
Network effects make the network exponentially more valuable with growing size
To develop a shared understanding is part of the research work because language underspecifies meaning: future ‘cloud’ research will build on it
And at the same time: linguistic relativity (Sapir-Whorf hypothesis): language culture restricts our thinking
The Idea
Spot unwanted fragmentation!
The goal of developing a recommender for flashmeeting is to use meta data ‐ to support researchers by pointing out other projects, researchers, or related topics they may not be aware of yet and that are closely related to their field of interest.
The Data
ECTEL
Meta-
Data
Flashmeeting
Meeting data is open (xml!)Complex database behind
we use rather small subset
The Method(s)
Degree Centralitynumber of (in/out) connections to others
Closenesshow close to all others
Betweennesshow often intermediary
Componentse.g. kmeans cluster (k=3)
(Social) Network Analysis (S/NA)
Meaningful Interaction Analysis (MIA)Making sense of latent-semantic networks.
The Defragmenter (1)
Pattern: from the co authorship network and ‐the co citations therein, a recommender ‐can identify when authors are working on the same topic (=keywords) but with different co authors and different literature. ‐
Intervention: propose to hold a ’get to know each other' Flashmeeting that may initiate desired defragmentation.
The Defragmenter (2)
Pattern: Communities are far from homogeneous. Sub-groups can emerge, particularly in big communities, which are connected by a small set (two or three) of members acting as bridge builders between otherwise disconnected components in the interaction graph.
Intervetion: Alerts about such structural dysfunctions including the provision of solutions such as joint virtual meetings can help to mend them and improve effective collaboration inside the global community.
First Results
Defrag meeting recommender
Spot unwanted fragmentatione.g. two authors work on the
same topic, but with different collaborator groups and with different literature
Intervention Instrument: automatically recommend
to hold a flashmeeting
Creating cohesion:defragment two groups
Communities are often not very dense, i.e. not resilient
With key persons withdrawing, the network can fragment
Recommend to build additional links, cutting out the middleman
More! FM Recommenders
• Group proposal recommendation: existing cliques can be discovered from graph components, recommending their members to form a group for supporting the management of joint meetings.
• Group closing recommendation: lack of activity in a group may indicate that it no longer exists as such. Confirming group disappearance would be necessary for keeping the server tidy and an accurate map of existing active communities.
• Group access recommender: when raising awareness about existing groups for a given individual, the participation of his/her contacts in a certain group is a strong indicator about the interest of the group for such a person. Recommendations for joining a given group based on contacts’ membership can help to avoid missing information.
• Meeting invitation recommender: awareness of community specific events can also be improved. Based on the known participants in the event as well as their contact relations, recommendations can be made for potential attendants.
Evaluation
Evolution-based evaluation
• The social network structure evolves in time• Compare recommendations based on
historical network data with links actually established (for a certain instant)
PROS: evaluation based on objective data CONS: lack of awareness (instead of non-relevance) can explain recommended connections not appearing in the real network
User-based evaluation
• Ask the user about the quality of the recommendations explicitly
• Questionnaire– Quantitative data (evaluation
metric)– Qualitative data (justification)
• Sample– Depends on actual recommendations
User-based evaluation
• PROS: – More accurate rewarding of
recommendations rising awareness – Deeper insight thanks to qualitative
information • CONS: – Missed links to recommend– Subjective information
(may be affected by other factors)– Data gathering– Statistical significance (sample size)
Structure-based evaluation
• Delete a sample of direct links and check if the system is able to rebuild the network, suggesting them as recommended collaborations.
• PROS: – Based on objective data
• CONS: – Deletions affect the
network structure
Our Preliminary Plan
• Use user survey:– ask for individual relevance ratings
of each recommendation (likert scale)– Stray in random distractors and use them as
a control group to test significance
Conclusion
Defragment today: http://fm.ea-tel.eu
Science requires networks.To understand, networks
need to communicate.With recommender systems unwanted
fragmentation can be spotted.
And interventions be scheduled.
Beware, the end is near.