When? - Collaborative Web Search: Who, What, Where, When, and Why - page 49

CHAPTER 5

When?

When a group collaborates on a search task, the temporal nature of their collaboration has important
implications for the types of algorithmic and user interface features that can best support them.
Broadly, there are two main temporal patterns to collaborative work: asynchronous and synchronous .
Synchronous collaboration refers to situations in which all group members are working at the same
time, whereas asynchronous collaboration refers to situations in which group members' efforts do not
necessarily overlap temporally. In this chapter, we provide an overview of systems designed to support
both methods of interaction, and look in-depth at S 3 ( Morris and Horvitz , 2007a ), which supports
asynchronous collaboration, and CoSense ( Paul and Morris , 2009 ), which supports synchronous
work.

5.1 ASYNCHRONOUS COLLABORATION

There are many situations in which collaborators work asynchronously on shared search tasks; this
is particularly true for tasks that may extend over multiple separate sessions, such as in complex
decision-making or research tasks like the one Martha and her family members engaged in. In the
absence of specialized support systems for collaborative search, the two primary strategies groups use
to coordinate their asynchronous search process are divide-and-conquer and brute force ( Morris, M.R. ,
2008 ).
In the divide-and-conquer approach, group members explicitly divide up a task, either ac-
cording to sub-tasks (e.g., Martha searched for price information for mold removal companies and
George for reputation information) or according to aspects of the search process, such as the search
sources used (e.g., Martha could have searched for mold removal companies via the Yellow Pages,
while George searched an online review site).
In contrast, the brute force approach reflects a lack of explicit division of labor. Group mem-
bers avoid the overhead of coordinating on their search strategies ahead of time, but they risk
realizing they have gathered redundant information when they share their findings. Martha and
Beth's collaboration had very low overhead, but it resulted in significant duplication of effort.
E-mailing a list of links to collaborators is the most common means of sharing the products of
an asynchronous searching session ( Morris, M.R. , 2008 ). Printing paper copies of relevant websites
to give to others is another common strategy. More effortful alternatives, such as creating a document
or webpage summarizing the findings of a search, are less prevalent. Sharing practices are listed in
Table 3.1.