Information Technology Reference
In-Depth Information
Only, Real World Only, Both, Neither, Neither but they
should be!}
Stats - presented detailed user statistics including comparisons among
users, a response to the ideas of beta users in the preliminary
study.
Once users had completed a given set of questions, they were presented with a
table comparing their answers with Facebook's data, and score and accuracy
metrics. As this page was generated, their results were archived into the dataset.
A Facebook application was chosen over other forms of study because it
automatically affords access to a real-time sampling of the participants' social
network that contains connections ranging from acquaintances to close friends.
While many likely connections can be inferred from other users' perceptions, an
important limitation of this study is that Facebook networks are inherently
incomplete pictures of one's social network. However, the quality and breadth of
already available data greatly simplifies establishing a picture of a participants'
social network, a key issue with previous real-world studies. Lewis et al. [4]
discuss these advantages and include: its inherent real world nature, full
population description, and complete demographic information. Simultaneously,
participant recruitment is not limited to a given social group, but rather can spread
organically outward as participants recommend the study to their friends, and can
be undertaken at any computer across the world.
Data points were selected to form points of comparison between users, evenly
split between self-ratings, those where the user rates their own connections,
present connections, and missing connections as summarized in pseudocode
below:
//Self-ratings
Select all current friends using application
Choose randomly among them if too many
Fill with random sample of friends if insufficient
//Present/Absent Connections (selected similarly)
Select all overlapping edges (other participants) where
connection present or absent
Order by whether have self-rated connection to either user in
edge
Add some random edges
//Selection
Seed a list evenly with three element groups
Select half to create random distribution
This algorithm optimizes overlap between both the participants' own connections
and with other participants.
3.2 User Survey
In order to refine the user experience and the effectiveness data selection
algorithm, a preliminary usability study was conducted consisting of a pre-use
questionnaire, exposure to the application, and a post-use questionnaire. The
initial application contained all basic features excepting the Stats page. Overall
Search WWH ::




Custom Search