NodeXLNetwork overview, discovery and exploration

for Microsoft Excel 2007http://www.codeplex.com/nodexl

Dan Fay (Microsoft Research - Redmond)Cody Dunne (U Maryland)

Marc Smith (Telligent)Vladimir Barash (MSR Silicon Valley/Cornell)

Tony Capone (Microsoft Research - Redmond)Natasa Milic-Frayling (Microsoft Research - Cambridge)

Eduarda Mendes Rodrigues (Microsoft Research - Cambridge)Eric Gleave (U Washington)Adam Perer (U Maryland)

Ben Shneiderman (U Maryland)

The NodeXL Team

Problem: No network chart in Excel

Problem: No network chart in

Excel

NodeXL: Network analysis and visualization tool

• Cyclic Graph data structures have limited support in existing Office tools

• Network analysis is of growing importance in academic, commercial, and Internet social media contexts

• Existing network analysis tools have command line interfaces or demand steep learning curves

• Many network data sets already live in Excel!

NodeXL: Goal: Make SNA easier

• Existing Social Network Tools are challenging for many novice users

• Tools like Excel are widely used• Leveraging a spreadsheet as a host for SNA

lowers barriers to network data analysis and display

Social Network Analysis ToolkitTools to support the study of the social network structure of

social media and other directed graph structures

Sociologist

“What are the structures of

communication in scientific

discussions?”

User Experience Information Visualization

“What are the best UI/UX workflows for network

analysis tools?”

Computer Scientist Algorithmicist for Social

Network Measures

“What are the measures and algorithms needed for understanding networks?”

8

The Ties that Blind?

9

Reply-To NetworkNetwork at distance 2 for the most prolific author of the microsoft.public.internetexplorer.general newsgroup

The Ties that Blind?

Darwin Bell11

Pajek without modification can sometimes reveal structures of great interest.

The Ties that Blind?

Two “answer people” with an emerging 3rd.

Mapping Newsgroup Social Ties

Microsoft.public.windowsxp.server.general13

14

• Answer person– Outward ties to local isolates– Relative absence of triangles– Few intense ties

• Reply Magnet– Ties from local isolates often

inward only– Sparse, few triangles– Few intense ties

Distinguishing attributes:

15

Distinguishing attributes:• Answer person

– Outward ties to local isolates– Relative absence of triangles– Few intense ties

• Discussion person– Ties from local isolates often

inward only– Dense, many triangles– Numerous intense ties

16

Clear and consistent signaturesof an “Answer Person”

• Light touch to numerous threads initiated by someone else

• Most ties are outward to local isolates• Many more ties to small fish than big fish

1

10

100

0 1 2 4 8 16 32 64

17

Roles Project• Using Netscan

data to derive social roles in Usenet

• Next steps: quantify & explore in more depth

Answer Person, microsoft.public.windows.server.general

Discussion, rec.kites

Flame, alt.flame

Social Support, alt.support.divorce

PUBLISHED in HICSS, JCMC, JoSS, IEEE Internet Communications (special issue on Social Networks) 18

NodeXL: Network Overview, Discovery and Exploration for Excel• Leverage spreadsheet for storage of edge and

vertex data

The NodeXL project is

Available via the CodePlex

Open Source Project

Hosting Site:

http://www.codeplex.com/nodexl

NodeXL workflow

data importation > processing > calculation > refinement > a network graph that tells a useful story

These steps include:

• Import data from several sources and file formats• Scrub data: Merge duplicate edges• Calculate network metrics• Insert sub-graph images• Auto-fill columns (and map data to display attributes):

- Set shape, color, opacity, size, and label/tooltip• Create clusters• Show graph• Read workbook• Adjust layout• Layout Again• Dynamic Filters – selectively hide edges and nodes

NodeXL: Import data from multiple sources:

• Multiple network “spigots” provide edge lists from several common sources and data formats.

Social media platforms are A source of multiple Social network data sets:

“Friends”“Replies”“Follows”“Comments”“Reads”“Co-edits”“Co-mentions”“Hybrids”

Export data to alternate file formats:

Prepare data for analysis

NodeXL: Import edges from other spreadsheets

• Map data columns from existing spreadsheets

NodeXL: Merge Duplicate Edges (if any)

• Aggregate duplicate edges and add a “Tie Strength Column” to store the count of “duplicates” (edges could be from multiple time slices).

NodeXL: Calculate Network Analytics and Metrics

• Starter library of basic network measures

• Users may unselect resource intensive measures

NodeXL: Insert network sub-graph images

• Create “ego-centric” networks for each node in the network

• Select number of degrees out to include

NodeXL:

Display nodes with subgraph images sorted by network attributes using Excel Data|Sort

NodeXL: Get reports of global network metrics

Metric Value

Graph Type Directed

Unique Edges 7,852

Edges With Duplicates 0

Total Edges 7,852

Self-Loops 10

Vertices 174

Graph Density 0.260514259

NodeXL: Display whole graph

• Toggle display of whole graph display pane with Show/Hide Graph Pane

NodeXL: Create a new whole graph display

• Select “Read Workbook” to load the graph into the Display Pane.

• The title “Document Actions” is imposed by Excel

NodeXL: Viewing the whole graph

NodeXL: Using Dynamic Filters to simplify the graph

• Each data column (including dates) associated with an edge or vertex is exposed with a slider filter.

• Filtered nodes and edges turn gray or become invisible

NodeXL: Apply dynamic filters to the data

NodeXL: Map data to display attributes

• Map Edge and Vertex attributes to size, width, color, opacity, and shape

NodeXL: Decorated Network Graphs

NodeXL: filtered, decorated, labeled networks

NodeXL: Clustered networks

NodeXL: Add URLS to Right-click menu of Nodes

NodeXL: Filtered clusters

Right click the canvass to control attributes of selected nodes

NodeXL: Import social networks from email

NodeXL: Import social networks from email

Systematic Yet Flexible Network Analysis Tasks1. Overall network metrics (e.g. number of nodes, number of

edges, density, diameter)2. Node rankings (e.g. degree, betweenness, closeness

centrality)3. Edge rankings (e.g. weight, betweenness centrality)4. Node rankings in pairs (e.g. degree vs. betweenness, plotted

on a scatter gram)5. Edge rankings in pairs6. Cohesive subgroups (e.g. finding communities in networks)7. Multiplexity (e.g. analyzing comparisons between different

edge types, such as friends vs. enemies)Shneiderman, Perer, Dunne

Micro ---- Scale ---- Macro

Micr

o ---

- Tim

e ---

- Mac

ro

Whole Graph

/ All Time

Sub Graph /

Narrow Time Slice

Sub Graph /

All Time

Whole Graph

/ Narrow

Time Slice

Filtering Network Diagrams

Random Layout

Fruchterman-Reingold Layout(Dense)

Fruchterman-Reingold Layout(Loose)

Random Layout (Decorated)

X = In-degree, Y = Out-degree

NodeXL Next Steps

• Enhanced layout controls– Smart selection of nodes

• Clustering and composite nodes– Add/remove a node to/from a cluster– Add/remove a node to/from a composite

• Add social network data sources:– Twitter, YouTube, Facebook, Outlook, Messenger,

etc.

NodeXL Partnerships and community

• University of Maryland• Northwestern University• Ohio University• Stanford University• University of Pennsylvania

7,000 + downloads on Codeplex

NodeXL User tasks and goals

NodeXLNetwork overview, discovery and exploration

for Microsoft Excel 2007http://www.codeplex.com/nodexl

Dan Fay (Microsoft Research - Redmond)Cody Dunne (U Maryland)

Marc Smith (Telligent)Vladimir Barash (MSR Silicon Valley/Cornell)

Tony Capone (Microsoft Research - Redmond)Natasa Milic-Frayling (Microsoft Research - Cambridge)

Eduarda Mendes Rodrigues (Microsoft Research - Cambridge)Eric Gleave (U Washington)Adam Perer (U Maryland)

Ben Shneiderman (U Maryland)