The 5 W's of Collaborative Search - CSAIL People

Document Sample
The 5 W's of Collaborative Search - CSAIL People Powered By Docstoc
					The 5 W’s of Collaborative Search

    Jaime Teevan and Merrie Morris
           Microsoft Research
Who, What, Where, When & Why?
•   Who does collaborative searches?
•   What are collaborative search tasks?
•   Where are the collaborators located?
•   When does the collaboration occur?     Frame

•   Why do people collaborate?
• How can we support collaborative search?
Who Does Collaborative Searches?
• Everyone!
  – More than 50% report collaborating for search
  – People rely on others even for individual search
• Consider type of relationship
• Privacy considerations
  – Friends v. strangers
  – Kids v. adults
 What Are Collaborative Search Tasks?
• Tend to be involved tasks (v. directed search)
• Some popular tasks:
  – Travel planning
  – Shopping
  – Technical information/literature
• Important: Important task be related to group
  – Work groups get benefit for work tasks
  – Social groups get benefit for social tasks
Where Are the Collaborators Located?
• Co-located
  – Ability to share control
  – Ability to partition task: Smart Splitting
• Remote
  – Awareness of others important
 Type 2 Diabetes Complications - American Diabetes Association
 Many of the complications of diabetes are strongly related to high blood sugar
 levels. It is believed that keeping your blood sugar levels in your target range is
 your best defense ...
 When Does the Collaboration Occur?
• Synchronous
  – Communication important
  – Support different roles
• Asynchronous
  – Need to preserve and expose state
  – Hard to identify future value of information
  – Can “collaborate” with future self
    Why Do People Collaborate?
• Group identification
  – Explicit                                          Task                         Friends

  – Implicit                                                            Age

                                                           Job team
• Group longevity                                                Job role
  – Task-based                                         Relevance judgments

  – Trait-based                                      Query selection      Interest group

                                                     Task-based               Trait-based
 How Can We Support Collaboration?
• Group identification
  – Match people into groups
  – Match task to appropriate group
• Algorithmic support for collaborative search
  – Identify documents of interest to group
  – Partition information appropriately to members
  – Summarize documents based on group interests
 How Can We Support Collaboration?
• Build a common evaluation framework
  – An example framework
     • Individual profiles
     • Grouping information
     • Relevance judgments for common queries
  – We have used the framework to study
     • Groupization
     • Smart splitting
     • Group hit highlighting
                        Thank you.
                          Jaime Teevan

Collaborative Search M. R. Morris and Teevan, J. Collaborative Search: Designing
for a Spectrum of Collaborative Styles. (Ed. G. Marchionini.) Morgan-Claypool 2009.

Groupization Teevan, J., M. R. Morris and S. Bush. Discovering and Using Groups
to Improve Personalized Search. WSDM 2009.

Exploring Solutions Morris, M. R., J. Teevan and S. Bush. Enhancing Collaborative
Web Search with Personalization: Groupization, Smart Splitting, and Group Hit-
Highlighting. CSCW 2008.
Group Types Morris, M. R. and J. Teevan. Understanding Groups' Properties as a
Means of Improving Collaborative Search Systems. JCDL Workshop on
Collaborative Information Retrieval 2008.

Shared By: