genSpace: Exploring Social Networking Metaphors for Scientific Collaborative Work Gail Kaiser, Swapneel Sheth, Chris Murphy {kaiser, swapneel, cs.columbia.edu Programming Systems Lab Department of Computer Science Columbia University

genSpace Goals Prior “collaboratories” for bioinformatics scientists  Tool sharing (e.g. BioCoRE from UIUC)‏BioCoRE  Data sharing (e.g. BSC from PNNL)‏BSC  Instead, we seek to introduce knowledge sharing  What tools / datasets should I use to investigate this problem?  Who do I know who also uses this tool / dataset?  Which tools and datasets work nicely together?  Where does this tool / dataset fit in a typical workflow?  When did I previously use this tool / dataset?  How can I get help (from an expert who is online right now)?

genSpace Goals Social Networking facilitates Collaborative Filtering  What movies would I like?  Who also likes this book?  Which food and wine go together?  Where does this song fit in the playlist?  When was this restaurant last reviewed?  How can I get help about ? We investigate Social Networking Models as an approach to Scientific Knowledge Sharing We are implementing a prototype for geWorkbenchgeWorkbench

genSpace Technical Details genSpace client is a geWorkbench “component”, separate centralized genSpace server Instrument geWorkbench to capture and record analysis events Aggregate event logs for communities of users Data-mine event patterns Automatically construct networks of “similar” users Automatically construct implicit workflows consisting of sequences of analysis tools Query via Instant Messaging (IM) “bots” Visualize networks and workflows

genSpace Architecture

genSpace plugin for geWorkbench

IM based query interface (jclaim)

Single Workflow Visualization

Workflow Relationship Visualization

genSpace Research Questions We do not aim to (directly) produce new results answering bioinformatics research questions easier Instead, we seek to make it easier for bioinformatics researchers to find answers to their own questions We do, however, seek to produce (direct) new research results of interest in software systems  Adapting a “social” metaphor for collaborative work  Data mining implicit workflows  Publish/subscribe based on implicit participation in tool and dataset use communities  Security/privacy implications and amelioration