Visualising Collaboration via Email: Finding the Key Players

Onn Azraai Puade, Theodor G Wyeld

( IEEE 2006)

Visualising Collaboration via Email: Finding the Key Players

Advisor ： Dr. Koh Jia-LingSpeaker ： Tai Yi-Ling

Date ： 2008.10.30

OutlineIntroduction

Collaboration Email Visualization

Case studyNetwork Diagram AnalysisEmail Content AnalysisVisualizing Collaboration Impact

Discussion

IntroductionEmail is an important form of asynchronous

communication.

Email often forms the backbone to research, industry, educational and other collaborations.

Visualizing analyses of email communication patterns during a collaborative activity help us better understand the nature of collaboration, and identify the key players.

IntroductionThis paper outlines a proof-of-concept prototype

collaborative email visualisation schema.

A new and novel method to identify the key players in a collaboration exercise based on their impact on the group.

It forms its conclusions based on how the individual players rate the importance of each other’s emails.

Case StudyThe collaboration involved the organization and

running of a workshop to develop resources for a multi-user game.

The workshop ran for three days. 20 individuals from 6 organizations were involved in the activity over 197 days.

The participants of ages 21-51 in this study came from diverse backgrounds.

Case StudyEmail as a communication tool was assumed.

The period chosen for analysis is just before and after the workshop was run.

There were 24 emails sent by 10 participants over this period.

Case StudyEach email included embedded prior emails,

subject descriptions, sender, receiver(s), date and message.

From this data, we were able to plot the connection between participants and the types of topics discussed.

Network Diagram AnalysisNetwork graphs were constructed from the

collection of emails.Node and link graphs were generated using

Pajek , a social network analysis visualization tool.

Undirected graphs:vertices -> emails Nodes -> participants

Network Diagram Analysis

Network Diagram Analysisadding email nodes

Email Content AnalysisAutomatic classification by data mining and

information retrieval techniques can be seen in many research studies.

To demonstrate this we recast Divitini and Farshchian’s [8] email roles as a classification system.

According to their content:A = AwarenessD = Decision makingE = Accessing expertF = FeedbackR = Resolving issues

Email Content Analysis

Email Content AnalysisConducting a survey with the participants

identified in the 24 email collection. To rate each individual email in term of its

importance on a scale of:0 – Not applicable1 – Not important2 – Important3 – Very important

Email Content AnalysisRe-organize the table by number of emails per

participant, type, ratings and average ratings

Email Content AnalysisAverage rating represents the ‘loudness’ (L) of a

participant’s message.

And multiply their loudness by the number of emails (N) sent, this is a measure of their overall ‘impact’ (I) on the collaboration.

L X N = I

Email Content Analysis

Email Content Analysis

Visualizing Collaboration ImpactThen visualize the results of these tabulations.Both these visualisations help us to gain ‘at a

glance’ a better understanding of the information contained in the tables.

It includes:Orange dot – number of mailBlue dot – loudnessDashed circle – impactConcentric ring – four rating scales

Visualizing Collaboration Impact

Visualizing Collaboration Impact

DiscussionFrom the visualisation of these two analyses,

there was more variation between participants in both loudness and impact.

There was little variation in loudness between the different types of email, But there was greater variation between impact than that displayed in the participant impact.

DiscussionPredefined roles, such as project leader,

manager, coordinator, and so on, do not necessarily generate the greatest impact on a collaborative project over time.

The traditional purpose of these roles – to make announcements on progress, meetings, and queries – is supported by the ‘by-type’ visualisation.