=Paper=
{{Paper
|id=Vol-3946/DARLI-AP_paper14
|storemode=property
|title=Epistemic Violence Against Female Artists and Scientists in Latin America on Wikipedia: Unveiling the Imbalance Between Minority and Majority Worlds using Graphs
|pdfUrl=https://ceur-ws.org/Vol-3946/DARLI-AP-14.pdf
|volume=Vol-3946
|authors=Genoveva Vargas-Solar,Alejandra Josiowicz
}}
==Epistemic Violence Against Female Artists and Scientists in Latin America on Wikipedia: Unveiling the Imbalance Between Minority and Majority Worlds using Graphs==
https://ceur-ws.org/Vol-3946/DARLI-AP-14.pdf
Epistemic Violence Against Female Artists and Scientists in Latin
America on Wikipedia: Unveiling the Imbalance Between
Minority and Majority Worlds using Graphs
Genoveva Vargas-Solar1 , Alejandra Josiowicz2
1
CNRS, Univ Lyon, INSA Lyon, UCBL, LIRIS, UMR5205, F-69221, France
2
Instituto de Letras, Universidade Estadual do Rio de Janeiro (ILE-UERJ) and Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de
Janeiro (FAPERJ), Brazil
Abstract
The democratization of the Internet and the perceived "universal" access to online content have long given the impression of progress
and inclusion. However, digital content overwhelmingly represents knowledge produced in English and within the majority world,
reflecting only a fraction of the knowledge created throughout history across diverse cultures. Epistemic violence remains pervasive in
much of the moderated content online, yet its extent is challenging to measure. This paper introduces a novel approach to address this
gap by proposing an Epistemic Violence Index applied to Wikipedia biographies of Latin American women scientists and writers. Our
study involves constructing a graph representation of the Wikipedia network connections for leading female figures in science and
literature from the 19th and 20th centuries. The analysis highlights their connections with influential voices both within the region
and in the majority world, evaluating the reciprocity and imbalance of these relationships. By leveraging these graphs, we compute
an Epistemic Violence Index based on an intersectional set of variables, including gender identity, socio-economic status, and race,
providing an initial step toward quantifying and addressing this persistent issue.
Keywords
Data science pipelines, graph analytics, Wikipedia, epistemic violence index
1. Introduction introduces the concept of Epistemic Violence and outlines a
quantitative approach for measuring it within Wikipedia’s
The democratization of the Internet and the perceived "uni- intellectual network. Section 4 presents an experimental
versal" access to online content have long given the impres- validation assessing the degree of epistemic violence in
sion of progress and inclusion. However, digital content Wikipedia articles on Latin American women intellectuals,
overwhelmingly represents knowledge produced in English comparing their network positioning to their connections.
and within the majority of the world, reflecting only a frac- Finally, Section 5 summarizes the findings and explores di-
tion of the knowledge created throughout history across rections for future research.
diverse cultures. Epistemic violence remains pervasive in
much of the moderated content online, yet its extent is chal-
lenging to measure. This paper introduces a novel approach 2. Related Work
to address this gap by proposing an Epistemic Violence In-
dex applied to Wikipedia biographies of Latin American The study of influential hubs in artistic and intellectual com-
women scientists and writers. munities lies at the intersection of digital humanities and
We analysed Wikipedia content across languages by cre- graph analytics. Researchers have employed computational
ating and exploring graphs to examine digital knowledge techniques and theoretical models to explore how certain
production and circulation inequalities. We drew on per- individuals or nodes in social networks act as pivotal points
spectives from Data Feminism and Digital Humanities in for knowledge dissemination, creative collaboration, and in-
the Global South. Our study involves constructing a graph tellectual influence. These studies combine network science,
representing the Wikipedia network connections of leading sociology, and computational humanities methods, provid-
female figures in science and literature from the 19th and ing insights into the dynamics of cultural and intellectual
20th centuries. The analysis highlights their connections ecosystems [1]. Thus, the potential contribution of network
with influential voices within the region and in the majority science to rethinking the dynamics of intellectual history
world, evaluating the reciprocity and imbalance of these in the humanities and social sciences in Latin America and
relationships. By leveraging these graphs, we compute an the world is considerable.
Epistemic Violence Index based on an intersectional set of
variables, including gender identity, socio-economic status, 2.1. Graph Analytics and Social Networks
and race. This provides an initial step toward quantifying
and addressing this persistent issue. Graph analytics, a key methodological approach in this field,
Accordingly, the remainder of this paper is structured as is widely used to model and analyze social networks. In
follows. Section 2 provides an overview of key studies on these networks, nodes represent individuals, and edges rep-
influential hubs within artistic and intellectual communi- resent social or professional relationships. Centrality mea-
ties and relevant graph analytics methodologies. Section 3 sures, such as betweenness centrality, eigenvector centrality,
and closeness centrality, are often employed to identify in-
Published in the Proceedings of the Workshops of the EDBT/ICDT 2025 fluential hubs.
Joint Conference (March 25-28, 2025), Barcelona, Spain One of the earliest theoretical contributions, Freeman’s
*
Genoveva Vargas-Solar. centrality metrics [2] laid the foundation for understanding
†
The authors’ list is alphabetical except for the first and second authors. the roles of nodes in social structures. Betweenness central-
$ genoveva.vargas-solar@cnrs.fr (G. Vargas-Solar);
alejandra.josiowicz@uerj.br (A. Josiowicz)
ity, in particular, has been critical for identifying individuals
© 2025 Copyright © 2025 for this paper by its authors. Use permitted under Creative Commons
License Attribution 4.0 International (CC BY 4.0).
CEUR
ceur-ws.org
Workshop ISSN 1613-0073
Proceedings
who act as “bridges” in artistic and intellectual communities, Co-authorship networks provide another lens for under-
connecting otherwise disparate subgroups. standing intellectual collaboration. Researchers like New-
The advent of Graph Neural Networks (GNNs) has en- man [9] demonstrated that scientific productivity and in-
hanced our ability to analyze complex social networks. Re- novation often emerge from highly connected hubs in co-
searchers such as Kipf and Welling [3] introduced methods authorship networks. These hubs publish prolifically and
for semi-supervised learning on graph structures, which bridge disciplines, fostering interdisciplinary knowledge
have been adapted for identifying hubs in artistic and aca- exchange [Newman, 2001].
demic communities. These methods can capture higher- Recent studies have revealed gender asymmetries in sci-
order relationships and provide richer representations of entific production in different fields and disciplines, arguing
nodes, revealing nuanced forms of influence. that there are persistent inequalities: women scientists are
under-represented globally in science citations [10], and
2.2. Digital Humanities and Social Networks data on their participation in specific scientific fields can be
challenging to find [11].
in Intellectual Communities The rise of platforms like ResearchGate and
The digital humanities field has embraced network analysis Academia.edu has allowed scholars to study intellec-
to study historical and contemporary cultural movements. tual influence in digital contexts. These platforms generate
Scholars use computational tools to analyze how collabo- large-scale datasets that can be analyzed using graph
ration and social dynamics shape intellectual and creative techniques to identify trending topics, influential authors,
outputs. and collaborative patterns[12].
A seminal project in this domain is the “Mapping the Re- Created in 2001, Wikipedia is a free, multilingual, open-
public of Letters” initiative by Stanford University. By ana- source encyclopedia edited and maintained by a community
lyzing correspondence between Enlightenment thinkers, re- of volunteer editors worldwide that has revolutionized the
searchers identified key figures, such as Voltaire and Diderot, creation and circulation of public knowledge. It plays an
as influential hubs facilitating the exchange of ideas across essential role in the dissemination of knowledge and in es-
Europe. This project demonstrated how network analysis tablishing avenues of dialogue between academia and the
could uncover the social infrastructure of intellectual move- general public. Recent studies on Wikipedia’s database of ar-
ments [4]. ticles and its community of editors have shown that they are
In contemporary contexts, projects like the “Art Markets” shaped by disparities in gender, language and geolocation
initiative use network analysis to map relationships between [13, 14, 15], which are accentuated outside English-speaking
artists, galleries, and collectors. These studies reveal how communities. However, there is a need for more research
a few prominent galleries or collectors often serve as hubs, on these dynamics in the case of languages such as Spanish
shaping artistic trends and market dynamics [5]. The influ- and Portuguese [16].
ence of these hubs is not merely economic but also extends In Latin America, intellectual dynamics are shaped by his-
to the promotion and visibility of specific artistic styles. torical hierarchies of knowledge production and circulation
From a digital humanities perspective, researchers like between centers and peripheries [17]. This is why studies
Klein and D’Ignazio, in their book Data Feminism [6], have on digital platforms such as Academia, ResearchGate and
explored how social and algorithmic biases impact the vis- Wikipedia in the Global South need to consider the back-
ibility of women and marginalized groups in intellectual ground of epistemic violence, data colonialism, and cultural
networks. Network analyses of Wikipedia or academic ci- domination that shape the region and its history [7].
tation graphs often reveal gendered patterns of influence
and invisibility, prompting calls for more inclusive digital 2.4. Crossovers Between Artistic and
archives.
Intellectual Networks
Perspectives on Digital Humanities from the Global South
have pointed to the inequalities in the production, distribu- Several studies bridge the gap between artistic and intellec-
tion and access to knowledge from the Global North and tual communities, highlighting their interconnected nature.
the Global South [7]. In particular, mapping and visualizing Cultural institutions like museums and universities often
intellectual networks with data science and digital humani- serve as meeting points for artistic and intellectual com-
ties tools can help to make visible the historical hierarchies munities. Network studies of these institutions reveal how
between privileged and marginalized lettered groups and they act as conduits for exchanging ideas. For example, the
think critically about how gaps and skews in intellectual Louvre and the Museum of Modern Art (MoMA) have been
history may be addressed. analyzed as influential hubs that connect artists, critics, and
academics[18].
2.3. Graph-Based Studies in Intellectual Platforms like X and Instagram have enabled the study of
real-time interactions in artistic and intellectual spheres.
Communities Researchers use graph analytics to track how hashtags,
The intellectual landscape, particularly academia, has been retweets, and mentions propagate through networks, iden-
extensively studied using graph-based techniques to identify tifying users or institutions that amplify discourse. Notably,
influential scholars and interdisciplinary connections. studies on #BlackLivesMatter and similar movements have
Studies on citation networks have long dominated this highlighted the role of influential nodes in shaping public
area. Metrics like PageRank, initially developed for web narratives [19]. Research on #BlackLivesMatter in particular
search engines, have been adapted to evaluate the influence has shown that male activists are overrepresented in users
of academic papers and authors. For instance, the works of referenced, which makes them more central in networks
Hirsch [8] on the h-index integrate network principles to in platforms such as X, pointing to the need to develop
quantify an individual’s academic centrality. intersectional frameworks to study anti-racist activism [20].
Studies on X have highlighted that referencing and nam- tuguese article on Carolina Maria de Jesus1 was created in
ing women and LGBTQIA+ intellectuals can articulate forms 2007, spans 35,7730 bytes, considered significant, and has
of resistance to the algorithmic hierarchies and infrastruc- undergone 411 edits. With 185 editors and 171,518 visits, it
tures of platforms such as X, making these figures more visi- is the most viewed among the 12 biographical pages about
ble and creating lineages that centre women and LGBTQIA+ the author in different languages.
individuals in renewed intellectual traditions [21]. The Portuguese article on Bertha Lutz 2 , on the other
hand, is rated as "good quality" by the platform’s automated
2.5. Discussion tool, which increases its visibility (it is worth noting that the
platform does not justify this category). Created in 2004, the
Despite significant advances, several challenges remain in article spans 40.096 bytes, has had 279 edits by 129 editors,
studying influential hubs in artistic and intellectual commu- and, with 333,449 visits, is also the most frequented page
nities. Historical studies often face limitations due to incom- about the author among the 24 languages of her biography.
plete or biased datasets. For example, archives may under- The Spanish article on Cecilia Grierson 3 spans 31 957
represent marginalized groups, leading to skewed analyses bytes, was created in 2006 and has undergone 592 edits by
[22]. As Klein and D’Ignazio [6] highlight, algorithms used 269 editors. It is also the most visited page on this intellec-
in network analysis can perpetuate existing biases. Ensur- tual among the 19 languages she has a biography on the
ing that the methodologies are inclusive and representative platform, with 826,328 visits.
remains an ongoing challenge. Meanwhile, the Spanish article on Silvina Ocampo 4 is
The intersection of graph analytics and digital human- the longest of the four, with 51 222 bytes. Created in 2006,
ities calls for interdisciplinary collaboration. Researchers it has received 925 edits and contributions from 333 editors,
must combine technical expertise in network science with making it the most edited and collaboratively built article
critical perspectives from the humanities to fully capture among the selection. It has had 683,580 visits in Spanish,
the complexities of influence. Most existing studies analyze the most visited among the 31 languages of her biography.
static networks, but intellectual and artistic communities
are inherently dynamic. Developing methods to analyze
3.2. Visibility and Invisibility of Intellectual
temporal changes and evolving hubs is a promising area of
research [23]. Women on Wikipedia
We created a graph of link connections between Wikipedia
biographies. In this graph, red edges link to the article, blue
3. Epistemic Violence edges link from it, and green edges link to biographies differ-
Epistemic violence refers to the systematic marginalization ent from the selected intellectual. Each biography’s bubble
and devaluation of knowledge, contributions, and perspec- size is proportional to each node’s degree of centrality. To
tives originating in the Global South [24], including Latin analyze these graphs, we must consider the networks of
American intellectuals, artists, and scientists. This phe- links and hyperlinks between articles that affect the visi-
nomenon often manifests as non-reciprocity in the networks bility of biographies. Specifically, if women’s articles are
of influence and acknowledgement between Global South poorly connected to other articles, they become more chal-
and Global North intellectuals. Latin American creators lenging to find. It is essential to clarify that creating links
are frequently excluded from global academic and cultural is a complex skill that involves identifying related pages
discourses, with their contributions either dismissed, ap- and adding hyperlinks to them, which has significant impli-
propriated, or underrepresented in citations, collaborations, cations for information retrieval and visibility on the Web.
and historical narratives [25]. For instance, connections Editors with less technical knowledge or experience may
between Latin American figures and their counterparts in be less aware of the importance of creating inbound and
the Global North are often asymmetrical, where the work outbound links and might make errors. This impacts the
of Global South intellectuals enriches or informs Northern structural centrality in the platform’s knowledge network,
projects without reciprocal acknowledgement or integration with consequences for the visibility of these intellectuals’
into canonical histories. This lack of reciprocity reinforces biographies. Such actions can either amplify or combat the
existing hierarchies, as Latin American contributions are existing inequalities and exclusions surrounding intellectual
treated as supplemental rather than foundational. In artistic women in Latin America.
and scientific circles, this marginalization is exacerbated by The graph shows that Carolina Maria de Jesus’s article is
structural barriers, including limited access to funding, in- relatively poorly connected compared to, for example, the
ternational publishing platforms, or exhibitions in globally Brazilian journalist Audálio Dantas. Dantas’s fame largely
recognized institutions. Such dynamics create fragmented stems from his role in first editing Carolina’s diary. Some im-
or invisibilized knowledge networks, perpetuating stereo- portant Brazilian literary writers referenced are Conceição
types of intellectual dependency while undermining the Evaristo, an acclaimed Afro-Brazilian writer whose article
autonomy and centrality of Latin American actors in shap- does not link back to Carolina Maria de Jesus, and Maria
ing global discourses. Firmina dos Reis, considered Brazil’s first Black novelist,
whose article also does not reference her. While the pres-
3.1. Latin American Intellectual Women on 1
https://xtools.wmcloud.org/articleinfo/pt.wikipedia.org/Carolina_
Wikipedia: Biographical Portraits Maria_de_Jesus?uselang=pt
2
https://xtools.wmcloud.org/articleinfo/pt.wikipedia.org/Bertha_
First, we visualized the general information about each au- Lutz?uselang=pt
thor’s article on their respective Wikipedia pages. The Por- 3
https://es.wikipedia.org/w/index.php?title=Cecilia_Grierson&
action=info
4
https://es.wikipedia.org/w/index.php?title=Silvina_Ocampo&action=
info
Figure 1: Carolina Maria de Jesus Connections Graph in Wikipedia
ence of these Afro-Brazilian authors is significant, the lack Global North peers:
of connection to other authors in the Brazilian literary canon (︃
InDegreeGS
)︃
further marginalizes and renders Carolina Maria de Jesus’s EVI = 𝛼 · 1−
OutDegreeGS
biography invisible, minimizing the impact her writing has
had on subsequent generations of Afro-Brazilian authors, + 𝛽 · (1 − ReciprocityGS )
women writers, and Brazilian literature overall. BetweennessGN − BetweennessGS
(︂ )︂
+𝛾·
Another important intellectual mentioned is Leonel BetweennessGN
Brizola, then-governor of Rio de Janeiro, who does not re-
(︃ )︃
EigenvectorCentralityGN − EigenvectorCentralityGS
+𝛿·
ciprocate with a reference to her. EigenvectorCentralityGN
These are important intellectuals. However, given Car- (︃ )︃
ClusteringGN − ClusteringGS
olina Maria de Jesus’s importance in Brazil’s intellectual +𝜖· .
ClusteringGN
and literary history, the network illustrates her lack of cen-
trality within the platform’s knowledge graph, relegating InDegree
(︁ )︁
• Indegree / outdegree 1 − OutDegreeGS Measures the
her to the margins of intellectual history in Brazil and Latin GS
ratio of incoming to outgoing references for the
America - a history that the platform has the potential to
Global South, highlighting the asymmetry where
map. This marginalization reveals the multiple layers of
outgoing links dominate over incoming ones.
oppression that intersect and compound as gender, race,
• Reciprocity = Number of Mutual Links
the proportion
class, and language interact and amplify their effects [26]. Total Number of Links
of bidirectional connections within the network.
It is also noteworthy that most authors and researchers
1 − 𝑅𝑒𝑐𝑖𝑝𝑟𝑜𝑐𝑖𝑡𝑦 Captures the lack of reciprocal
mentioned in the article as studying Carolina’s life and work
connections.
are not linked because they lack Wikipedia articles, further
• Betweenness centrality measures the Global South
deepening the marginalization of her work’s study.
intellectual’s bridging role compared to Global North
intellectuals.
(︁ )︁
3.3. Towards an Epistemic Violence Index • Betweenness GN −BetweennessGS
BetweennessGN
Normalizes marginaliza-
Epistemic Violence Index (EVI) quantifies the degree of tion in bridging positions.
asymmetric visibility, marginalization, and lack of reci- • Eigenvector Centrality measures influence disparity.
EigenvectorCentralityGN −EigenvectorCentralityGS
(︁ )︁
procity for an intellectual in the Global South relative to • EigenvectorCentralityGN
normalizes
influence asymmetry.
• Clustering Coefficient compares network integra- network, comparing the PageRank of Global South intellec-
tion
(︁ for Global South )︁ vs. Global North intellectuals. tuals to those of the Global North can highlight disparities
ClusteringGN −ClusteringGS
• ClusteringGN
normalizes exclusion from in perceived authority or significance.
tightly connected communities. Similarly, the clustering coefficient, which measures how
well a node’s neighbors are interconnected, provides in-
The EVI combines graph metrics as follows: sights into network integration. A low clustering coefficient
for Global South intellectuals might indicate their exclusion
• Visibility Disparity: Ratio of in-degree centrality to from tightly-knit intellectual or cultural cliques that are
out-degree centrality. In-degree centrality measures more prevalent in Global North networks.
the number of incoming links to an intellectual’s Metrics like structural holes and brokerage roles, such as
Wikipedia page. A low in-degree compared to peers constraint or effective size, further reveal whether Global
in the Global North indicates a lack of references South intellectuals occupy marginalized positions in the net-
and visibility. Out-degree centrality measures the work, limiting their ability to bridge diverse sub-networks
number of outgoing links from Wikipedia to other or access critical connections.
articles. A high out-degree with a low in-degree Lastly, analyzing centralization differences can shed light
indicates that the intellectual’s work is referenced on broader epistemic inequalities: a highly centralized struc-
outward but not reciprocally acknowledged. Asym- ture dominated by Global North intellectuals underscores
metry, the ratio of in-degree to out-degree, can re- their pivotal role in the network, reinforcing the peripheral
veal whether the Global South intellectual’s work status of Global South figures. Together, these metrics offer
contributes more to the network (out-degree) than a comprehensive framework to evaluate and understand
they receive recognition for (in-degree). structural disparities in knowledge networks.
• Reciprocity Disparity: Proportion of unreciprocated
links for the Global South intellectual.
• Positional Marginalization: Difference in eigenvec- 4. Example: Application of the
tor centrality and betweenness centrality between Epistemic Violence Index (EVI) to
the Global South intellectual and an average Global
North intellectual. Betweenness centrality measures
Wikipedia Networks
how often a node acts as a bridge between other This section demonstrates how we utilized the Epistemic
nodes in the graph. A Global South intellectual with Violence Index (EVI) to estimate epistemic violence between
low betweenness compared to peers in the Global two intellectual networks with differing centrality. Our
North would indicate their marginalization in con- analysis compares intellectuals from the Global North and
necting intellectual or cultural sub-networks, signi- Global South.
fying epistemic exclusion. The Eigenvector central-
ity measures the influence of a node based on the
importance of its neighbors. A low eigenvector cen- 4.1. EVI for comparing pairs of intellectuals
trality for Global South intellectuals would highlight Consider two pairs of intellectuals, one from the Global
their marginal role in a broader, high-impact intel- North (Marie Curie and Virginia Woolf) and one from the
lectual network dominated by Global North nodes. Global South (Carolina Maria de Jesus and Silvina Ocampo).
Reciprocity measures the proportion of mutual links All have Wikipedia pages with varying levels of connectivity
between nodes. For example, if the page of a Global and recognition in the global knowledge network. We aim
South intellectual links to many Global North intel- to compute their Epistemic Violence Index (EVI) to evaluate
lectuals but receives few or no reciprocal links, it structural disparities in their representation and visibility.
demonstrates asymmetric visibility and epistemic The computed EVI score for Carolina Maria de Jesus is
inequality. 0.65, which indicates a significant disparity in representa-
• Network Exclusion: Clustering coefficient dispar- tion and influence compared to Marie Curie. This score
ity and absence from dense intellectual communi- highlights the structural marginalization of Global South
ties. The clustering coefficient measures how well a intellectuals within the Wikipedia network. It reflects issues
node’s neighbours are connected. A low clustering like fewer incoming links (visibility), lower reciprocity, and
coefficient for Global South intellectuals might indi- reduced centrality measures, all contributing to epistemic
cate their exclusion from tightly-knit intellectual or inequities.
cultural cliques common in Global North networks. This EVI score (0, 59) indicates a significant disparity in
Silvina Ocampo’s representation and influence within the
Higher EVI values indicate greater epistemic violence, re-
network compared to Virginia Woolf. The higher disparities
flecting significant asymmetries in visibility, reciprocity, and
in reciprocity, centrality measures, and clustering highlight
marginalization in intellectual networks. A well-connected
structural marginalization in the network.
Global North intellectual with reciprocal and central connec-
We further test EVI to analyze how it behaves with a se-
tions would have a lower EVI. In contrast, a Global South in-
lection of nodes from the graph mapping Carolina Maria de
tellectual with poor reciprocity, low influence, and marginal-
Jesus’s network of links. For this task, we employ different
ization would have a higher EVI.
techniques of normalization. For visibility disparity and
reciprocity disparity, we compute the empirical distribution
Comparing Majority and Minority Worlds Intellec- of the data and use the associated cumulative distribution
tual’s Graphs. PageRank is a key metric used to assess the function to obtain a uniform normalization between 0 and 1.
relative importance of a node based on the quality and quan- For positional marginalization and network exclusion, we
tity of its incoming links. When applied to the Wikipedia rescale the values linearly. We present preliminary results
In-degree Out-degree Reciprocity Betweeness Eigenvector Clutering
mutual links out of total links centrality centrality coefficient
Marie Curie
150 80 70% 0,45 0,62 0,58
Maria Carolina de Jesus 30 60 40% 0,12 0,15 0,2
Visibility Reciprocity Betweeness centrality Eigenvector centrality Clustering coefficient
disparity Disparity disparity disparity disparity
0,5 60% 0,733333333 0,758064516 0,655172414
Epistemic violence index
EVI 0,64931405
Figure 2: Epistemic violence index for Maria Carolina de Jesus compared with Marie Curie
In-degree Out-degree Reciprocity Betweeness Eigenvector Clutering
mutual links out of total links centrality centrality coefficient
Virgina Woolf
150 100 75% 0,6 0,8 0,65
Silvina Ocampo 40 60 35% 0,25 0,2 0,22
Visibility Reciprocity Betweeness centrality Eigenvector centrality Clustering coefficient
disparity Disparity disparity disparity disparity
0,33333333 65% 0,583333333 0,75 0,661538462
Epistemic violence index
EVI 0,59564103
Figure 3: Epistemic violence index for Silvina Ocampo compared with Virginia Woolf
for a selection of nodes linking to and from De Jesus’ article • Conceição Evaristo (0,4337)
(see Table in figure 4). • Mariana Crioula (0,4452)
The table presents Maria Carolina de Jesus’s EVI scores
with various intellectuals and cultural figures, providing in- These intellectuals and figures have significantly lower EVI
sights into the extent of epistemic violence she experiences values, indicating that Maria Carolina de Jesus has minimal
in connection with them. The EVI values represent the level visibility within their network. This suggests that her con-
of epistemic marginalization, with higher values indicating tributions are less acknowledged in their biographical links
greater epistemic violence (i.e., lower reciprocity, visibility, or references, reinforcing epistemic exclusion.
and influence in the knowledge network).
Lowest EVI (Less Epistemic Violence, Higher Visibility) Key Observations Marginalization in Literary and Politi-
cal Networks: Notably, key figures in Brazilian literature like
• Euriclides de Jesus Zerbini (0,5413) Machado de Assis or Brazilian politics like Leonel Brizola
• Maria Firmina dos Reis (0,5445) have lower connectivity with Maria Carolina de Jesus, sug-
• Solano Trindade (0,5440) gesting a lack of recognition of her contributions within the
• Zenaide Zen (0,5456) mainstream cultural and historical intellectual canon. It is
worth noting that despite being Afro-descendant, Machado
These figures have the highest network reciprocity with de Assis has long been located in the Brazilian literary canon,
Maria Carolina de Jesus, suggesting their Wikipedia pages which explains his low level of reciprocity with Maria Car-
reference her more frequently, reducing her epistemic invis- olina de Jesus.
ibility. Higher Reciprocity in Black and Marginalized Intellectual Net-
Moderate EVI (Moderate Epistemic Violence) works: Figures like Solano Trindade and Maria Firmina dos
• Audálio Dantas (0,5318) Reis, both prominent Afro-Brazilian intellectuals and writ-
ers, exhibit higher reciprocity, implying that Maria Carolina
• Marielle Franco (0,5110)
de Jesus is more visible in networks that focus on racial and
• Soul Ra (0,5220)
cultural resistance.
• Tia Ciata (0,4987) Audálio Dantas’s Role in Visibility: Despite Dantas’s instru-
These connections exhibit some degree of epistemic in- mental role in publishing Carolina’s work, his EVI score
equality but have relatively balanced visibility and refer- remains moderate (0.5318), suggesting a limited reciprocal
ences compared to Maria Carolina de Jesus. acknowledgement of her intellectual contributions.
Highest EVI (Most Epistemic Violence, Lower Visibility) The EVI scores highlight the structural epistemic violence
against Maria Carolina de Jesus, particularly in mainstream
• Machado de Assis (0,3456) literary and political networks, where her work is underrep-
• Leonel Brizola (0,4535) resented. However, she has higher recognition in Black in-
Figure 4: EVI of Carolina Maria De Jesus’ node and her connections
tellectual circles, indicating that her contributions are more
acknowledged within historically marginalized knowledge
communities. Addressing these disparities requires efforts
to interconnect her legacy with wider intellectual traditions.
4.2. Comparing communities of
intellectuals
The EVI is now calculated as a weighted average of six dis-
tinct measures, each designed to capture specific aspects of
network dynamics: visibility disparity, lack of reciprocity,
marginalization, lack of influence, exclusion from tightly
knit subgroups, and overall lack of connections. These mea-
sures are derived from appropriately normalized standard
centrality indices, such as degree centrality, betweenness
centrality, Eigenvector centrality, and clustering coefficient,
all scaled to the unit interval. This normalization ensures
that the measures are comparable and can be combined into
a single composite index.
In this section, we build on the previous definition of the
EVI by using network statistics to assign an EVI value to
each node in the graph of interest, rather than comparing
pairs of nodes from the Global North and South. This shift
simplifies the interpretation of the EVI, making it more in-
tuitive and directly applicable to the analysis of a specific
network. By focusing on individual nodes, the EVI pro-
vides a clearer understanding of each node’s position and
influence within the network. Figure 5: EVI Values of Carolina Maria de Jesus’ Contacts
In this updated version (see Table 5), we introduce an
additional component based on total degree centrality. This
component directly captures the overall centrality of a node,
reflecting whether it occupies a central or peripheral posi- of their connections.
tion within the intellectual network under consideration. This method of calculating the EVI offers a nuanced yet
This enhancement ensures that the EVI not only accounts straightforward understanding of network dynamics. By
for the nuanced aspects of network dynamics but also pro- focusing on individual nodes, it highlights disparities in vis-
vides a straightforward measure of a node’s prominence ibility and influence. The inclusion of total degree centrality
based on its total connections. Together, these improve- ensures the EVI reflects not only the number of connec-
ments make the EVI a more robust and interpretable tool tions but also each node’s overall centrality. This makes
for analyzing network equity and visibility. the EVI a valuable tool for identifying key influencers and
The table offers a snapshot of individuals’ visibility and in- marginalized individuals within intellectual or social net-
fluence within a network, quantified by the EVI. This index works, offering insights that can guide strategies to enhance
combines various network statistics, emphasizing both the equity and visibility.
quantity and quality of connections. Higher EVI values in-
dicate more central and influential individuals, while lower
values suggest marginalization or peripheral positions. This
5. Conclusions and Future Work
approach simplifies the interpretation of network dynamics Advancements in graph analytics and the digital humani-
and underscores disparities in visibility and influence. ties have benefited the study of influential hubs in artistic
The EVI values in the table range from 0.267843 (Solano and intellectual communities. From historical analyses of
Trindade) to 1.0 (José Correia Leite). Individuals with lower Enlightenment thinkers to contemporary studies of social
EVI values, such as Solano Trindade and Porfírio da Paz, are media interactions, researchers have unveiled the pivotal
likely marginalized or peripheral, potentially having fewer roles that certain individuals and institutions play in shaping
connections, less influence, or exclusion from tightly knit knowledge and creativity. However, future research will be
subgroups. Conversely, individuals with higher EVI values, crucial to address challenges such as databases, algorithmic
like José Correia Leite and João Cândido, are more central fairness, and dynamic modelling.
and influential, likely possessing more connections, greater The Epistemic Violence Index (EVI) offers a practical
influence, and integration into tightly knit subgroups. The framework for addressing disparities in knowledge represen-
EVI thus provides a comprehensive measure of an individ- tation. From a policy perspective, this analysis can inform
ual’s network position, capturing both the extent and quality
the development of initiatives to enhance the visibility of academic social network sites, Frontiers in Digital
Global South intellectuals by creating more effective inter- Humanities 6 (2019) 5.
linking strategies within platforms like Wikipedia. The EVI [13] X. Zeng, J. Chen, E. Yan, C. Ni, Gender and
score is a powerful quantitative measure to evaluate the country biases in wikipedia citations to scholarly
impact of systemic biases, enabling researchers and policy- publications, Journal of the Association for Infor-
makers to identify areas where representation gaps persist. mation Science and Technology 74 (2023) 219–233.
Additionally, advocacy groups can leverage the EVI to raise URL: https://asistdl.onlinelibrary.wiley.com/doi/epdf/
awareness about these inequities and advocate for more eq- 10.1002/asi.24723, accessed on June 4, 2024.
uitable representation in digital knowledge spaces, fostering [14] W. Luo, J. Adams, H. Brueckner, The ladies vanish?
a more inclusive and balanced intellectual landscape. american sociology and the genealogy of its missing
By integrating computational techniques with humanistic women on wikipedia, Comparative Sociology 17 (2018)
inquiry, this field will continue to deepen our understanding 519–556. doi:10.1163/15691330-12341471.
of the social fabric of intellectual and artistic life. [15] J. J. B.-V. Núria Ferran-Ferrer, J. Minguillón, Wikipedia
gender gap: A scoping review, Profesional De La in-
formación Information Professional 32 (2023). doi:10.
6. Acknowledgements 3145/epi.2023.nov.17.
[16] P. Beytía, C. Wagner, Visibility layers: a framework
The work reported in this paper is partially funded by
for systematising the gender gap in wikipedia content,
the collaboration between the UERJ and the CNRS, the
Internet Policy Review (2022). doi:10.14763/2022.
project FRIENDLY 5 funded by the LIRIS laboratory LIRIS
1.1621.
intergroup program, Lyon, and the FAPERJ project Nº E-
[17] F. Beigel, El proyecto de ciencia abierta en un mundo
26/200.247/2023.
desigual, Relaciones Internacionales (2022) 163–181.
[18] P. Bourdieu, The Field of Cultural Production: Es-
References says on Art and Literature, Columbia University Press,
1993.
[1] R. Ahnert, S. E. Ahnert, C. N. Coleman, S. B. Wein- [19] S. J. Jackson, M. Bailey, B. F. Welles, HashtagActivism:
gart, The network turn: Changing perspectives in the Networks of Race and Gender Justice, MIT Press, 2020.
humanities, Cambridge University Press, 2020. [20] D. Freelon, C. McIlwain, M. Clark, Beyond the hash-
[2] L. C. Freeman, Centrality in social networks concep- tags:# ferguson,# blacklivesmatter, and the online
tual clarification, Social Networks 1 (1979) 215–239. struggle for offline justice (research report), Center
[3] T. N. Kipf, M. Welling, Semi-supervised classification for Media & Social Impact. http://cmsimpact. org/blm-
with graph convolutional networks, in: International report (2016).
Conference on Learning Representations (ICLR), 2017. [21] A. J. Josiowicz, M. Méndez, Decolonial antiracist fem-
[4] S. University, Mapping the republic of letters, 2020. inist digital activism: Naming carolina maria de je-
URL: https://republicofletters.stanford.edu/. sus, lélia gonzález, and marielle franco on twitter, in:
[5] O. Velthuis, Talking Prices: Symbolic Meanings of Black Lives Matter in Latin America: Continuities in
Prices on the Market for Contemporary Art, Princeton Racism, Cross-National Resistance and Mobilization
University Press, 2011. in the Americas, Springer, 2024, pp. 409–440.
[6] L. F. Klein, C. D’Ignazio, Data Feminism, MIT Press, [22] M. Terras, Digitisation and digital resources in the
2020. humanities, Annual Review of Information Science
[7] P. Ricaurte, S. Chaudhuri, D. Fiormonte, Global De- and Technology 45 (2011) 115–145.
bates in the Digital Humanities, University of Min- [23] P. Holme, J. Saramäki, Temporal networks, Physics
nesota Press, Minneapolis, 2022. Reports 519 (2012) 97–125.
[8] J. E. Hirsch, An index to quantify an individual’s [24] G. C. Spivak, Can the subaltern speak?, Die
scientific research output, Proceedings of the National Philosophin 14 (1988) 42–58. doi:10.5840/
Academy of Sciences 102 (2005) 16569–16572. philosophin200314275.
[9] M. E. J. Newman, The structure of scientific collabora- [25] S. C. Ricaurte, Paola, D. Fiormonte, Global Debates in
tion networks, Proceedings of the National Academy the Digital Humanities, University of Minnesota Press,
of Sciences 98 (2001) 404–409. 2022.
[10] L. R. Ruiz, A. A. R. León, Índice de continuidad y [26] P. H. Collins, M. Andersen, Race, Class & Gender: An
contemporaneidad (icc): medir, visualizar y significar Anthology, 9th ed., Cengage Learning, Boston, 2016.
las asimetrías de género en la producción científica
sobre el trastorno de la personalidad, revista hispana
para el analisis de redes sociales (2023). doi:https:
//doi.org/10.5565/rev/redes.998.
[11] G. VARGAS-SOLAR, Intersectional study of the gen-
der gap in stemthrough the identification of miss-
ing datasets aboutwomen: A multisided problem,
Appl. Sci. (2022). doi:https://doi.org/10.3390/
app12125813.
[12] K. Jordan, From social networks to publishing plat-
forms: A review of the history and scholarship of
5
http://vargas-solar.com/friendly/