Networks: Gephi (and a little bit of Palladio)

```NETWORKS: GEPHI (AND A
@DJWrisley
#RRSI2014, UTSC, May 2014
NETWORKS
THEORY, ANALYSIS,
VISUALIZATION
Network analysis – a term encompassing a wide
variety of practices with applications throughout
(social) scientific and digital humanistic domains
 “Network theory concerns itself with the study of
graphs as a representation of either symmetric
relations or, more generally, of asymmetric
relations between discrete objects.” (Wikipedia, 6
italicized words are mine—all debatable in
humanist circles)
 Network theory has its own conceptual
vocabulary to express relationships between
objects (e.g. betweenness, centrality, density,
path length, modularity) – how can we interpret
these analytical terms for humanities data?

NETWORK THEORY, ANALYSIS,
VISUALIZATION (2)
Social network analysis SNA looks at
relationships between actors – what is the nature
of interaction?
 Latour adds objects in actor-network theory ANT
– what could the relation of an actor and an
object be?
 Networks once drawn (drawn by MorettiNetwork Theory, Plot Analysis), are now digitally
created and manipulated
 Powerful way of exploring multidimensional
multi-scalar data (Brughmans)

SOME BEGINNING REMARKS
@ELIJAHMEEKS ON THE NETWORK

"The network is not a social network or
geographic network or logical network but rather
a primitive object capable of and useful for the
modeling and analysis of relationships between a
wide variety of objects."
<https://dhs.stanford.edu/visualization/morenetworks/>
BASIC PRINCIPLES (1)




Not all study of networks is quantitative (Brughmans
on Malkin, 2011), just as every mapping is not made
of spatial data onto a map interface
Digital tools for network visualization and analysis
use tabular data (quantification can be a challenge,
Digital tools like Gephi allow both for networks to be
explored visually, and for static visuals of them to be
exported. It does not allow for sharing.
Data visualization is a kind of “problem-posing”; we
should avoid fetishizing the final visual.
(McCosker/Wilke) – “diagrammatic” thinking
BASIC PRINCIPLES (2)

Tabular data used by network viz platforms are
of two basic sorts
Nodes (discrete entities in a network, and any fixed
 Edges (specific instances of relations between nodes)

NB: Gephi does generate a nodes table if it is missing
(option: create missing nodes)
DIFFERENT TOOLS
NETWORK VISUALIZATION TOOLS

Gephi (standalone)







NetXL
Sci²



Elijah Meeks
https://dhs.stanford.edu/gephi-workshop/
https://dhs.stanford.edu/visualization/more-networks/*
Non proprietary code – plug ins
Scott Weingart
Indiana MOOC on visualization
Points of comparison: performance, usability, filter,
formats of data ownership, data portability, cost
EXAMPLE 1
NB: Examples Increase in Scale
ONE TEXT
RAMON LLULL, BOOK OF THE LOVER AND THE BELOVED (LATE
13TH) HTTP://WWW.AM.UB.EDU/~JMIRALDA/LLULLTRA.HTML
LLULL DATA SNAPSHOT (@DJWRISLEY
@TRACEY_DH) NODES (LEFT) EDGES (RIGHT)
LLULL’S “SOCIAL” NETWORKS
(@DJWRISLEY @TRACEY_DH)
LLULL “SOCIAL” NETWORKS, MINUS THE
NARRATOR (@DJWRISLEY @TRACEY_DH)
ALGORITHMIC LLULL
Fruchterman
Force Atlas
TINKER TAILOR SOLDIER SPY(@HAYAY44)
TINKER TAILOR, “THE KARLA COLLECTION”
(@HAYAY44)
14TH C MEDIEVAL FRENCH (AROUND JEAN DE
MEUN) – STYLOMETRIC ANALYSIS OF 70+ TEXTS
J RICHARDS (WUPPERTAL – GERMANY)
STYLOMETRIC DATA SAMPLE

Open file RRSI stylo experiment
250+ TEXTS OF MEDIEVAL FRENCH – TEST
CASE FOR POWER OF STYLOMETRY GIVEN
DIALECTAL DIFFERENCE (@DJWRISLEY AND
RICHARDS)
Parts of gephi interface
 Visualizing the network using the same file

J
@GRANDJEANMARTIN
NETWORKS OF NEWSPAPERS 19TH C VIA
@RYANCORDELL
IMAGE: ORBIS: THE STANFORD
GEOSPATIAL NETWORK MODEL OF THE
ROMAN WORLD
SPACES AND TEXTS – A NETWORK WITH
GEPHI @DJWRISLEY
VIE DE SAINT LOUIS TEXT - PLACE
@DJWRISLEY
JERUSALEM – ASPECTED BY FORM
@DJWRISLEY
PAUL’S EXAMPLES
A BRIEF COMPARISON Gephi







standalone
open source
Established large user
community = many
plug-ins
Takes a while to learn
Science/humanities use
Not easy to share
without giving away
data
Network statistics








web-based
free
Relatively new
Learning curve low
Specifically for the
humanities
simple to deal with data
multi-dimensionality
Data is not kept
“without any barriers”