FLOSS Communication Centralization Plot, Exponentially Weighted

Created: 2009-02-07 18:50:44

Download Workflow

The analysis in this workflow represents the basis of the analysis in our paper, Social dynamics of FLOSS team communication across channels. This workflow uses WSDL components to select periodized data from the FLOSSmole database and generate sociomatrices. The workflow parses the threaded list structure into a communication network based on reply-to relationships. In the analysis process, an edge weighting is applied so that older messages receive less weight using an exponential decay function; this is intended to moderate the effects of using a sliding window of observations for dynamic analysis. The weighted sociomatrices are then dichotomized according to a threshold, and their centralities are calculated using Râ€™s sna package, and plotted in a time series. The final output demonstrates communication centralization trends over time in a FLOSS project.

The analysis workflow allows users to set parameter values for the size of the sliding window, the date ranges covered, and the communication venues analyzed. Data for the analysis in the paper examined dynamics in different communication channels; the data sets included up to 90 periods (with a 90-day sliding window) and 10Kâ€™s of email messages or forum posts. Date periods are currently only in months, and exception handling for empty periods is implemented both in the WSDL components and the RShell scripts. There are a couple of inelegant solutions involving temporary output files, which are configured based on a standard Mac file structure. Future development effort could focus on solving these issues and on optimizing performance.

Preview

Download as scalable diagram (SVG)

Run

Run this Workflow in the Taverna Workbench...

Option 1:

Copy and paste this link into File > 'Open workflow location...'
http://myexperiment.org/workflows/144/download?version=4
[ More Info Expand ]

Workflow Components

Inputs (4)

Name	Description
Start_Date	Starting date for the time period of analysis.
End_Date	End date for the time period of analysis. Must be entered as yyyy-mm-dd.
Project_list	Names of project venues to include in this analysis. Specifying multiple projects is additive. Project names with available data as of 2/15/08 include: fire-tracker-feature-requests fire-tracker-translations fire-talk fire-core fire-development gaim-cabal gaim-devel gaim-forum-users-helping-users gaim-tracker-plugins gaim-tracker-patches gaim-tracker-smileys gaim-tracker-rej-patches
sliding_window	Size of the sliding window: 0 (for no sliding window), 1 for a 60-day sliding window, 2 for a 90-day sliding window, etc.

Processors (22)

Name	Type	Description
step	stringconstant	Specification of the amount of time covered in each sliding window unit. Currently the default is a month (30 days) and no other step sizes are supported.
Centralization_Plot	rshell	Plots a time series of network centralizations with some basic measures of central tendency included in the graph labels.
Calculate_Centralization	rshell	Calculates the network centralization for each network matrix.
Extract_period_end	local	Shim to extract period end date from event.
Extract_From	local	Shim to extract the sender from the event.
Extract_period_start	local	Shim to extract the starting date for the event.
delist_dates	local	Shim to change from a list object into a text object.
Extract_To	local	Shim to extract recepient for each event.
delist_period_end	local	Shim to change from a list object into a text object.
MergeProjectNames	local	Shim to concatenate project names for inclusion in the labeling on the centralization time series plot.
delist_to	local	Shim to change from a list object into a text object.
Extract_event_date	local	Shim to extract the event date from the event.
delist_from	local	Shim to change from a list object into a text object.
delist_event_date	local	Shim to change from a list object into a text object.
Split_to_single_events	local	Shim to split each set of events returned by the web service into a list of events.
delist_period_start	local	Shim to change from a list object into a text object.
Remove_duplicate_strings	local	Shim to remove duplicates.
Calculate_weight	rshell	Function to calculate an exponentially decayed edge weight based on event recency.
split_periods	local	Shim to split the periods returned from the web service into a list of periods.
EventsForProjectsInPeriod	arbitrarywsdl	Web service that retrieves the venue events for the dates in the time period from the FLOSSEventBrowser database.
MatrixBuilderR	arbitrarywsdl	Web service to turn a pair list and weights into a sociomatrix suited for use in R.
GetPeriods	arbitrarywsdl	Web service that calculates the period dates for the time series, taking into account the sliding window.

Beanshells (0)

Outputs (2)

Name	Description
PNGoutput	Time series plot of the centralizations for the communication networks.
centralizations	Text output of the dates and network centralizations, for use in other applications.

Links (33)

Source	Sink
End_Date	GetPeriods:end_date
Calculate_Centralization:output	Centralization_Plot:cent_list
Extract_From:nodelist	delist_from:stringlist
Extract_To:nodelist	delist_to:stringlist
Extract_event_date:nodelist	delist_event_date:stringlist
Extract_period_end:nodelist	delist_period_end:stringlist
Extract_period_start:nodelist	delist_period_start:stringlist
GetPeriods:return	split_periods:input
MatrixBuilderR:return	Calculate_Centralization:filecontents
Project_list	EventsForProjectsInPeriod:project_list
Remove_duplicate_strings:strippedlist	delist_dates:stringlist
Split_to_single_events:nodelistAsXML	Extract_From:xml-text
Split_to_single_events:nodelistAsXML	Extract_To:xml-text
Split_to_single_events:nodelistAsXML	Extract_event_date:xml-text
Split_to_single_events:nodelistAsXML	Extract_period_end:xml-text
Start_Date	GetPeriods:start_date
Calculate_weight:weight	MatrixBuilderR:weight_list
Project_list	MergeProjectNames:stringlist
MergeProjectNames:concatenated	Centralization_Plot:project_name
Split_to_single_events:nodelistAsXML	Extract_period_start:xml-text
delist_dates:concatenated	Centralization_Plot:date_list
delist_event_date:concatenated	Calculate_weight:event_datetime
delist_from:concatenated	MatrixBuilderR:from_list
delist_period_end:concatenated	Calculate_weight:end_date
delist_period_end:concatenated	Remove_duplicate_strings:stringlist
delist_period_start:concatenated	Calculate_weight:first_date
delist_to:concatenated	MatrixBuilderR:to_list
sliding_window	GetPeriods:extend_window_back_by
EventsForProjectsInPeriod:return	Split_to_single_events:xml-text
split_periods:return	EventsForProjectsInPeriod:period_xml
step:value	GetPeriods:step
Centralization_Plot:centralizations	centralizations
Centralization_Plot:g	PNGoutput

Coordinations (0)

Information Workflow Type

Taverna 1

Information Uploader

Andrea Wiggins

Information License

All versions of this Workflow are licensed under:

Information Version 4 (latest) (of 4)

Information Credits (3)

(People/Groups)

Information Attributions (0)

(Workflows/Files)

None

Information Tags (9)

Uploader tags

Log in to add Tags

Information Shared with Groups (1)

Free/Libre Open Source Software

Information Featured In Packs (2)

Log in to add to one of your Packs

Information Attributed By (1)

(Workflows/Files)

Output from FLOSS Communication Centralization Workflow

Information Favourited By (0)

No one

Information Statistics

7678 viewings

3303 downloads

[ see breakdown ]

Citations (0)

None

Version History

In chronological order:

FLOSS Communication Centralization Plot

Created by Andrea Wiggins on Sunday 10 February 2008 01:15:28 (UTC)

Last edited by Andrea Wiggins on Sunday 10 February 2008 01:41:21 (UTC)
FLOSS Communication Centralization Plot, Exponentially Weighted

Created by Andrea Wiggins on Sunday 10 February 2008 01:15:28 (UTC)

Last edited by Andrea Wiggins on Friday 15 February 2008 20:02:47 (UTC)

Revision comment:

This revision includes a bit more workflow metadata, the most important of which is that the description for the Project_list input now contains a list of the project data available to use with this workflow as of 2/15/2008
FLOSS Communication Centralization Plot, Exponentially Weighted

Created by Andrea Wiggins on Sunday 10 February 2008 01:15:28 (UTC)

Last edited by Andrea Wiggins on Friday 07 March 2008 01:55:29 (UTC)

Revision comment:

This revision eliminates two ungainly text output files. The ~/temp.data file, previously used to write out and read back in a sociomatrix, has been replaced with an internalized read.csv(textConnection()) call. The date-centralization pairlist that was previously written to ~/Desktop/cent-values.txt is now a proper workflow output.
FLOSS Communication Centralization Plot, Exponentially Weighted

Created by Andrea Wiggins on Saturday 07 February 2009 18:50:44 (UTC)

Revision comment:

Updated the workflow to include full metadata.

Reviews (0)

No reviews yet

Be the first to review!

Comments (2)

View Timeline

Log in to make a comment

Andrea Wiggins

We have made this workflow available to use with some FLOSSmole data. There is a list of the project names in the Project_list input metadata description that contains the data that are available for use as of 2/15/2008.

The project names are:

fire-tracker-feature-requests
fire-tracker-translations
fire-talk
fire-core
fire-development
gaim-cabal
gaim-devel
gaim-forum-users-helping-users
gaim-tracker-plugins
gaim-tracker-patches
gaim-tracker-smileys
gaim-tracker-rej-patches

Andrea Wiggins

Start and end dates are also required for using this workflow. The "active" project date ranges that we have chosen for the projects are:

Fire, 2001-08-16 (registry) through 2006-03-15 (one month following final release 1.5.6)

Gaim, 1999-11-13 (registry) through 2007-04-24 (when the project changed name to Pidgin and its lists moved)

Note that not all date ranges contain data for all venues; some were not active as early as others and some were ad-hoc communication venues (i.e. gaim-cabal).

Other workflows that use similar services (1)

Taverna 1

Uploader

Andrea Wiggins

FLOSS Communication Centralization Plot, U... (2)

Download

Created: 2009-02-07

Credits: Andrea Wiggins Crowston James Howison