Received: 6 February 2018 | Revised: 23 March 2018 | Accepted: 25 March 2018

DOI: 10.1111/tgis.12342

RESEARCH ARTICLE

The limits of GIS: Towards a GIS of place

Alberto Giordano1 | Tim Cole2

1
Department of Geography, Texas State
University, San Marcos, Texas
Abstract
2
Department of History, University of
In this article we reflect back on our decade-long collaboration on the
Bristol, Bristol, United Kingdom geographies of the Holocaust to argue for a GIS of place. Our previ-
ous work on ghettoization in Budapest and on the spatio-temporal
Correspondence
patterns of Jewish persecution in Italy had a marked spatial dimen-
Alberto Giordano, Department of
Geography, Texas State University, sion, both in the research questions we set out to answer and the
San Marcos, TX 78666. methods we used, which were largely quantitative. During the course
Email: [email protected]
of our research, we progressively came to realize that a spatial per-
Funding information spective favors the voice of the perpetrator and that to fully
National Science Foundation; National comprehend and understand the geography of the Holocaust, we
Endowment for the Humanities; Toni Schiff
Memorial Fund needed to engage with the voice of the victim, extend the set of
methods and tools used, and broaden our epistemology. While pro-
posing a fully-fledged model of a qualitative GIS of the places and
spaces of the Holocaust is beyond the scope of this article, we: (a)
argue for the integration of social network analysis, corpus linguistics,
and spatio-temporal methods and for a mixed-methods analytical
approach and (b) note how the topological and relational foundations
we identify as fundamental to a GIS of place parallel the long-
standing call for an “integrated history” of the Holocaust.

1 | INTRODUCTION

Reflecting on the limits of GIS is not a new exercise. In the late 1980s and early 1990s, both the representation model
of GIS and its worldview—who did GIS benefit?—came under scrutiny by geographers writing under the broad umbrella
of “critical GIS.” (See Schuurman (2000) and, more recently, O’Sullivan (2006) and Thatcher et al. (2016) for a summary
and an historical overview of the debate.) Early NCGIA Technical Papers from this period explored topics that include
the language of spatial relation, cognitive science and GIS, and public participation GIS, while seminal works by Pickles
(1995) and others (Curry, 1998; Obermeyer, 1995; Sheppard, 1995) questioned the ideological foundations of the field.
As a response from within the field, several academic GIS authors—for the debate barely registered outside of
universities—attempted to re-envision GIS from alternative perspectives, including Kwan (2002) from a feminist one.
Especially relevant to our objectives in this article is the discourse around qualitative GIS (Cope & Elwood, 2009)—
which is largely internal to geography—and, in a much broader context, the increased use of computational technolo-
gies and big data methods in the digital humanities and the spatial humanities (Bodenhamer, Corrigan, & Harris, 2010;
Dear, Ketchum, Luria, & Richardson, 2011; Drucker, 2012; Travis, 2015).

Transactions in GIS. 2018;1–13. wileyonlinelibrary.com/journal/tgis V

However, relevant developments took place well beyond initiatives rooted in the spatial humanities. Emerging in
the context of computer science and artificial intelligence (AI) as an extension of qualitative reasoning (QSR), qualitative
spatial reasoning is concerned with representing continuous properties of the world by discrete systems of symbols
(Cohn & Hazarika, 2001). Cohn and Renz (2007) outline the major axes of QSR, including the choice of a representa-
tional formalism, the ontology of spatial entities, and the definition of primitive relations and operators. As Cohn and
Hazarika (2001, p. 2) note, being multidimensional, space is not adequately represented by a single scalar quantity. The
goal of QSR is then to design representations that can answer qualitative queries without much numerical information,
and to derive and manipulate qualitative spatial representations that efficiently and correctly abstract important spatial
aspects of the underlying data (Bailey Kellog & Zhao, 2004). It is important to note that these research questions—in
particular the focus and concerns on spatial ontologies and the search for a framework for formal representations—are
not new in geography and GIScience, but refer back to the early days of GIScience and the pioneering work of David
Mark and others on spatial cognition and language (see, e.g., Egenhofer & Mark, 1995; Frank & Mark, 1991; Smith &
Mark, 2001). What is new is the integration of AI methods and tools in the context of geography, and for the specific
representation and analysis of qualitative spaces. Of particular importance given our own interests is how qualitative
spatial and temporal representation and reasoning (QSTR) extends QSR to incorporate a temporal element (for a brief
€n, 2017).
review, see Klippel & Wallgru
Another relevant area of development, that we return to below, is the intersection of corpus linguistics and GIS,
seen for example in the work of Ian Gregory and others at Lancaster on geographical textual analysis, and especially in
the Corpus of Lake District Writing project, which focuses on place names and the identification of geographical fea-
tures such as waterfalls, woodland, or farms (Donaldson, Gregory, & Taylor, 2017; Gregory, Cooper, Hardie, & Rayson,
2015; Gregory & Donaldson, 2016; Rayson et al., 2017). Similar questions are tackled by Kim, Vasardani, and Winter
(2017), who focus on the question of resolving ambiguous place names by exploring their relations to other spatial fea-
tures, and Song et al. (2017), who developed and tested a method for detecting and extracting vague cognitive
regions.
With a somewhat broader perspective (i.e., with the objective of providing an analytical and conceptual frame-
work) we have the idea of geo-narratives as developed by Kwan and Ding (2008), as a distinctive qualitative GIS
approach, and Mennis, Mason, and Cao’s (2013) work on combining qualitative and quantitative data and methodolo-
gies within the context of GIS, using visualization as the means of inquiry. For a more explicitly cartographic perspec-
tive, see the idea of deep mapping and spatial narratives (Bodenhamer, Corrigan, & Harris, 2015), inductive
visualization (Knowles, Westerveld, & Strom, 2015), and, from an ethnographic perspective, Kawano, Munaim, Goto,
Shobugawa, and Naito (2016) look at how “rich media”—maps, photography, video composition, big data, and interac-
tive web platforms—can serve as representations to communicate the study of spaces and urban life.
While recognizing the validity and relevance of the issues raised by these and other authors, and the overall posi-
tive contribution they have made to the academic field of GIScience, here we want to focus on one aspect, namely the
relationship that exists—or does not exist—between GIS and place. While place has been a major subject of geographi-
cal research for decades, the subject matter of GIS, the methods of GIScience, and much of the research in the field
have concentrated and revolved around the concept of space, intended as an abstract container framed by the Carte-
sian grid of scientific cartography (Goodchild, 2004). When it comes to representing social relations and the dynamics
of social power, which are the basis for one theory of place that may be particularly relevant to developing a GIS of
place (Lefebvre, 1974/1991), the spatial perspective of GIS appears lacking, a point certainly not lost in the debates on
“critical GIS.” As other researchers do, Pavlovskaya (2017, p. 5437) links qualitative GIS with the emergence of the digi-
tal humanities and the geospatial humanities, noting how the “field of geohumanities . . . focuses on the meaning of
place”.
In this article, we argue for a GIS of place that combines the quantitative perspective of spatial analysis with quali-
tative methods and data, and we see as our research agenda for the future the building of a bridge between “tradi-
tional” GIS—a “GIS of space”—and a “GIS of place”—more specifically, a GIS of place and space. Our starting point is a
multi-year collaborative work on the geographies of the Holocaust, as it is while studying the Holocaust that we have
GIORDANO AND COLE | 3

become convinced of the necessity to integrate and enrich the analytical powers of GIS with other types of data and
voices. Such an integrated framework has a correspondent in Holocaust scholarship. As Saul Friedländer pointed out in
the first of his two-volume history of Nazi Germany and the Jews, “Establishing a historical account of the Holocaust in
which the policies of the perpetrators, the attitudes of surrounding society, and the world of the victims could be
addressed within an integrated framework remains a major challenge” (Friedländer, 1997, p. 1). Friedländer’s call for
integrated histories reflected broader concerns with the way that the historical literature on the Holocaust had tended
to follow separate paths. On the one hand, an ever-growing literature—building on the ground-breaking work of Raul
Hilberg (Hilberg, 1961)—drew on the paperwork of the perpetrators to understand both how and why the Nazi German
state implemented a policy of persecution and ultimately mass murder of European Jews. However, the voices of those
persecuted and murdered were almost entirely absent from this burgeoning historiography. Bringing the voices of the
victims together with the paperwork of the perpetrators was, for Friedländer, more than simply ethical necessity. It was
also a way to better narrate the complex history (and geography) of the Holocaust (Friedländer, 1997, p. 2).
Rather than an “integrated” history—or an “integrated” geography—being a seamless blending of these two sour-
ces, bringing them together in “integrated” methodologies, texts reveal key tensions in experience and understanding
place. As Paul Jaskot has illustrated in examining the architecture of Auschwitz, “the structures that served as sites for
the persecution of the European Jews were designed by the SS, built by forced laborers, and experienced by camp
inmates; each category of participant bore a particular relationship to each building. The tension between these varied
meanings needs to be reflected in any integrated history of the Holocaust” (Jaskot, 2015, p. 4). For Jaskot, “it is only
by thinking relationally that we do not lose the survivor’s story in the vast abstraction of a systemic analysis of the con-
centration camp network. Each records its side of the history, each remains distinct, and each is necessary to explaining
the Holocaust” (Jaskot, 2015, p. 5). The idea of thinking relationally is one that we return to later in this article as it is
key, we argue, to thinking about a GIS of place as “occupied space” that foregrounds topology (Bodenhamer et al.,
2015, p. 2). Before we outline such perspectives, we briefly introduce earlier work in working with GIS methods to
uncover the spatiality of genocide during the implementation of the Holocaust in Italy and Hungary.

2 | EXAMPLES FROM ITALY AND BUDAPEST

The Historical GIS of the Holocaust in Italy (HGIS Italy) (Giordano & Holian, 2014) includes the names of 6,116 victims
for which the location of arrest was known, along with information regarding their Last name, First name, Place of birth,
Nationality—Italian or foreign, Date of birth, Mother, Father, Spouse, Place of arrest, Date of arrest, Place of internment in
Italy, Convoy of deportation, Date of deportation, Destination camp, and Fate. The database also includes the Nationality
of the perpetrator—Italians, Germans, or Italians with Germans—the Last place of residence of the victim before the arrest,
and any additional Place(s) of internment of the victim in Italy. The starting point for the construction of the HGIS was a
list of approximately 9,000 Jews deported from Italy during the Holocaust (Picciotto Fargion, 1991/2002).
The HGIS allows for the representation and exploration of the spatial patterns of the Holocaust in Italy, which
have not simply a pedagogical value but also a value as a research tool. A summary of these patterns is presented in
Figure 1. The solid circles on the map indicate the number of arrests in each place and portray the Holocaust at the
local scale: we can see the ubiquity of the event, the fact that arrests were conducted in all parts of northern and cen-
tral Italy, and in centers small and large. The number of arrests was especially high in traditional centers of Jewish life
in Italy—like Rome, Florence, Venice, and Trieste—and in places where the contingencies of the war found an unusual
concentration of victims—like Borgo San Dalmazzo (Jews escaping from France) or the area north of Milan (Jews
escaping from Italy into neutral Switzerland). The five clusters in gray—generated through a k-means clustering tech-
nique—provide an aggregate regional view of the local patterns.
As with any aggregation, the patterns need to be contextualized and taken with a pinch of salt. The four clusters
in northern Italy can be explained and are not unexpected: this is the case for the smaller clusters in northwestern Italy
centered in Turin, Genoa, and Borgo San Dalmazzo; around Milan and the Swiss border area; and around Florence.
These small clusters correspond to cities of traditional Jewish presence (Turin, Genoa, and Milan), and places where
4 | GIORDANO AND COLE

FIGURE 1 Spatio-temporal patterns of the Holocaust in Italy

Jews were arrested in large numbers (Borgo San Dalmazzo and the Swiss border area). The largest and elongated clus-
ter in northeastern Italy includes some traditional Jewish centers and corresponds to the orientation of the principal
axis of transportation across the Po plain. The fifth cluster—the area from Rome to Civitella del Tronto—simply reflects
the location of these two places with respect to each other, with no major centers of arrest in between, and does not
therefore constitute a meaningful cluster, one that can be explained and interpreted—and perhaps produce new knowl-
edge—by the known characteristics of the Holocaust in Italy. (The two locations are separated by the Apennine Moun-
tains, a barrier that slowed at the time—and still slows to some extent—physical communications between the two
sides.) An additional step toward exploring the spatial patterns of the Holocaust in Italy at different scales (e.g., at the
scale of the city, or the province) is to add interactivity to it, allowing viewers to explore the dataset on their own. This
is the objective of the online geovisualization “Arrests of Italian Jews, 1943–1945” (https://web.stanford.edu/group/
spatialhistory/cgi-bin/site/viz.php?id5383&project_id50), created by Erik Steiner, Alberto Giordano, and others. In
addition to offering interactivity and a possibility of exploring the dataset at different scales and by selected variables
associated with the victim, a timeline at the bottom of the geovisualization allows viewers to explore the spatio-
temporal patterns of the Holocaust in Italy, thus allowing for a chronological perspective that a single static map is
unable to offer.
The HGIS of the Budapest Ghetto (HGIS Budapest) (Cole & Giordano, 2014a) differs from the Italian HGIS in
important ways. First, it is at the urban scale; and second, it is a GIS of buildings and not people. Created from a variety
of primary sources—the Italian HGIS relied on secondary sources—and superimposed on a map of Budapest c. 1944,
this HGIS shows the location of over 3,300 individual residences and over 100 public places. Identified by an address,
these individual buildings were designated for residences and patronage for the Jewish population of Budapest (close
to 200,000 at the time) during the second half of 1944 (see Figure 2). Mapping the various stages of the ghettoization
process in Budapest allowed us to highlight its spatial and temporal patterns at different scales of analysis, from the
building to the district, for Buda and for Pest, and then for the whole city. Thus, we determined that the location of
Jewish-designated residences corresponded for the most part to the traditional geography of Jewish Budapest, thus
leading to the assertion that the perpetrators in Budapest “brought the ghetto to the Jews” rather than the other way
around. We also determined that while the reconfigurations of the ghetto over time were substantially different from
GIORDANO AND COLE | 5

FIGURE 2 Budapest Ghetto

earlier plans, these successive implementation stages (see Figure 2) did not significantly alter the level of spatial con-
centration of residences, and, by extension, the concentration of the victimized population (Cole & Giordano, 2014a).
Network analysis allowed us to focus on the scale of the street and on people’s movement from their place of res-
idence to key public places, including market halls for buying food, hospitals for receiving care, and, crucial to the
establishment of the so-called “International Ghetto,” the location of the Swedish and Swiss legations for acquiring pro-
tective papers. What emerged clearly from the analysis were the differential constraints that Jews lived under because
of the importance of physical distance from people and resources. Similar to what we did for Italy, an interactive ani-
mation showing Jews walking to market halls, with a commentary from the authors, adds interactivity to the static
map and the ability for the user to explore the geography of food access (https://web.stanford.edu/group/spatialhis-
tory/cgi-bin/site/viz.php?id5411&project_id50). It is by looking at the maps, static and dynamic, that we hypothesized
that Jews living in District V were subject to more potential competition for food there than in other parts of the city,
while suburban market halls in Buda seemed better positioned to serve the Jewish population than market halls in
Pest. Similarly, locational advantages emerged for access to the offices of the Swedish legations.
Overall, network analysis revealed the existence of “invisible walls,” which we defined as the result of the potential
for accessing places and people key to survival; this, in a condition of almost total control of the victimized population,
was largely determined by location (Cole & Giordano, 2014b; Giordano & Cole, 2011). However, and here is one of
the key moments that led us to rethink and redefine the potential of GIS and GIScience for studying the Holocaust,
these “invisible walls” could be breached by socio-economic status, and the existence of pre-ghettoization social net-
works of support. This is clear not only from testimonies but also, for example, by examining the lists of some of the
Jews admitted into the “International Ghetto”—a “privileged” ghetto, in the sense that it was envisioned as the waysta-
tion to emigration rather than deportation—as a result of the acquisition of papers from the Swedish legation: a major-
ity of people on the lists we examined lived quite far from the legation and were instead from areas of the city near
the “International Ghetto” and in general from more affluent neighborhoods (Cole, 2016; Cole & Giordano, 2014b).
6 | GIORDANO AND COLE

3 | THE LIMITS OF GIS REPRESENTATION

As the example above from Budapest’s “International Ghetto” makes clear, spatial barriers such as the physical distance
between home and the Swedish legation buildings where paperwork was given out could be overcome by friendships,
contacts, and in general social networks that predated the implementation of the Nazi-era ghettos. What is especially
relevant is that these networks of social relations occur in place—in the case of the Budapest Ghetto, the building and
the apartment within the building. So, for example, interviewing Jewish survivor Judit Brody, when asked whether it
mattered that the family was able to stay within their own apartment, she replied that staying put meant “everything.”
It was not only the importance of remaining in a physical and material place, but also the significance of remaining in a
social place that mattered (Cole & Giordano, 2014b).
The importance of place as a site of social relations also emerged in the testimony of one of the 6,116 victims
recorded in HGIS Italy: Gilberto Salmoni. Arrested in April 1944 and deported to Buchenwald in August 1944 with his
brother, Gilberto’s story is, in a sense, a mirror of Judit’s: whereas Judit stayed in Budapest, and actively tried to “stay
put” to take advantage of pre-existing social networks, Gilberto and his family left their hometown, were arrested while
trying to escape to Switzerland, were moved to several prisons and a camp (Fossoli) in Italy, and were finally deported
to Buchenwald (Gilberto and his brother) and Auschwitz (Gilberto’s mother, father, and sister) (see Figure 3 for the
entire family trajectory). Gilberto’s narrative is one that alternates movement between places with, once at a place (e.g.,
Buchenwald), the identification of a place within the place (e.g., a certain barrack) that maximized the chances of sur-
vival, and that is where he and his brother—for the limited degree of agency they were able to retain—“stayed put.”
Here again, the social networks he and his brother were able to create in place played a key role as a strategy of sur-
vival. In Gilberto’s testimony, the camp, the sub-camp, the Appelplatz (roll call square), the barrack, the half-barrack, the

FIGURE 3 The Salmoni family of Italy

table he and other prisoners had dinner at, and the bunk bed, were not only where different things happened, but
even more importantly they were the loci of social relations, which appear to be highly spatialized and are indeed
explicitly discussed by Gilberto as happening not only “in place,” but explicitly “of place,” in the sense that they could
only occur in those specific places and times (Giordano’s interview with Gilberto Salmoni, January 8, 2017; see also
Salmoni, 2016).
Reflecting back on the HGIS of Budapest and Italy, we became progressively convinced that the tool of GIS and
the methods of GIScience are especially useful to tell the story of the Holocaust from the perspective of the perpetra-
tor, for example through mapping out where in Italy arrests took place, or where in Budapest Jews were concentrated
into ghetto houses. This does not mean that GIS and GIScience are intrinsically, inherently, or by design tools of—and
for—oppression. On the contrary, GIS and GIScience can uncover critical aspects of the geography of oppression by
showing—as we did for Budapest and Italy—how widespread and systematic the spatio-temporal patterns of victimiza-
tion were, at different scales and for different victims. What we hope to have shown in earlier work is the utility of
GIS and GIScience to study not only the Holocaust, but also other instances of genocide (Burleson & Giordano, 2016).
In doing this we relied on spatial data with well-defined and measurable levels of thematic, temporal, and positional
accuracy, and that have little or no intrinsic uncertainty; this is, we dare say, especially important when dealing with
ethically charged topics like the Holocaust and genocide, particularly given the specific context of Holocaust denial
which foregrounds the importance of accuracy—and objective truthfulness—in historical narratives. The HGIS applica-
tions we have built include records of arrest, lists of buildings and public places, dates and places and names. All, of
course, have their own accuracy issues, but these can generally be measured and estimated.
However, the fact is that a GIS of place, one that gives voice to the experiences of the victims, requires a radical
conceptual rethinking of the worldview and data model of traditional GIS. A GIS of place cannot, in our opinion, simply
consist of narratives (in text or video) superimposed on the spatial abstract view of a GIS application; good examples
of this approach exist, but are not what we are aiming for here (Aitken, 2015; Mennis et al., 2013; Watts, 2010). Such
approaches do not enable the kind of truly “integrated” or “relational” approach that scholars of the Holocaust have
urged are necessary. Superimposing narratives onto a GIS map is significantly different from integrating them into—
and with—the spatial analysis. For example, in order to explore place and social networks as key strategies of survival,
we need to, metaphorically at least, enter into the buildings and the camps and listen to the voices of those who lived
there. This requires new sources (testimonies and memoirs), new digital methods (appropriate for studying testimo-
nies), and a shift from spatial quantitative methods to spatial qualitative methods—or rather, the integration of quanti-
tative and qualitative methods into a mixed-methods analysis.
In the Introduction, we reviewed a series of approaches and methods—including QSR, geo-narratives, deep map-
ping, and others—that, in the context of a mixed-methods analytical approach, may contribute to the theory and prac-
tice of a GIS of place. Here we want to focus on one such approach and method—corpus linguistics, which we have
used in the past. Corpus linguistics methods are useful to parse large oral history collections to identify key terms that
include indicators of place characteristics such as “neighbor,” “corridor,” “courtyard,” “apartment,” and so on, as this is
where social relations occur. The real question then becomes how to integrate the powerful spatial–analytical capabil-
ities of GIS, which lead to the unearthing of patterns as shown in the preceding section, at the same time being able to
represent the voice of the victims, operationally defined as spatial strategies of survival that are the effect and result of
the deployment of social relations in place. We realize that this is a rather limited view of what a GIS of place may look
like, but unless the task is delimited and carefully defined, the risk is vagueness and ineffectuality. The point where the
policies of the perpetrator meet and interact with the victim is at the resolution of the individual. It is here, we think,
that the abstract view of space—the perpetrator’s policies—meets the place-making activities of victims as well as of
perpetrators (e.g., a camp guard and a prisoner).
To explain and contextualize what we mean, it is important to note that in geography and GIScience the concept
of resolution takes a precisely delimited and narrow meaning. At its most fundamental, resolution refers to the smallest
measurement unit at which data are collected and/or analyzed. In GIScience as well as in geography, resolution has
three components—one spatial, one temporal, and one thematic. Thus, the spatial resolution of HGIS Budapest is the
8 | GIORDANO AND COLE

address of the building, the temporal resolution is the period of time the building was a designated Jewish residence,
and the thematic resolution is the building itself. For Italy, the spatial resolution is the city or town where a victim was
arrested, its temporal resolution is the day, and its thematic resolution (what we are collecting data on and measuring)
is the individual victim. From the individual building or victim, it is possible to “scale up” to study the spatio-temporal
patterns of the Holocaust in Italy at the city, regional, and national scale and, for Budapest, at the neighborhood, dis-
trict, and urban scale. At every scale, however, the resolution remained that of the individual building or person.
Gilberto Salmoni’s strategies of survival, as already noted, were explicitly spatial and social, and consisted in alter-
nate moments of movement and stasis and the construction and reconstruction of social networks explicitly dependent
on—and through—place. In Buchenwald, for example, Gilberto realized that the camp was in constant movement, being
reshaped continually as prisoners were reassigned to other barracks or to other sub-camps, or were sent out on con-
stantly shifting labor assignments. As mentioned earlier, the camp, the sub-camp, the Appelplatz, the barrack block, the
half-barrack, the bunk bed, and the table where he sat with other prisoners to eat dinner were not only where different
things happened, but even more importantly they were the loci of spatially (and temporally) specific social relations. By
reducing the immensity and mystery of the Buchenwald system to the smallest and most manageable geographical
units—scale after scale—Gilberto managed to make sense of, relate to, and survive in the camp. This richness of experi-
ence can be represented with GIS only formally and symbolically. In order to capture the more nuanced elements of
Gilberto’s spatial narrative of the Holocaust, we need to develop techniques that deal specifically with the written
word, with the aim of bringing the two approaches together, in an “integrated” and “relational” way. As discussed
briefly in the Introduction, we see potential for QSR in this context.

4 | FOR A GIS OF PLACE

Building a GIS of place presents conceptual as well as practical and functional challenges. It is, first of all, a collaborative
enterprise, as we envision many possible implementations of GISs of place, rather than a monolithic and general-
purpose system. That is why we circumscribed it in the preceding section as a set of tools and methods to study the
deployment of social relations in place. For this view of place, we are indebted to the work of Yi-Fu Tuan, who, in his
classic study of space and place, observed that “home” is not necessarily anchored in place (Tuan, 1977): it can reside
in a person, or a personal relationship, such as the deeply comforting bond that can exist between parent and child, or
any two people who trust and rely upon one another. In other words, place and social relations are inextricably inter-
twined and one cannot be studied without the other (the social is spatial and the spatial is social). A research agenda on
a GIS of place in the context of the Holocaust is one that answers questions such as: how are the places of the Holo-
caust formed? what kind of relationships shape and define these places? what happens in these places? how are these
places connected? And, crucial not only to the Holocaust but to genocide studies as well, how does this all relate to
human survival and resilience?
A GIS of place will require a rethinking and extension of the analytical tools and methods of GIScience toward a
mixed-methods approach, rather than a purely, or even prevalently, quantitative approach, and an explicit link and con-
versation with the methods and approaches of the digital humanities and the spatial humanities. For an example of
this conversation, in a different context and with different objectives, see the idea of “deep maps” as visual spatial nar-
ratives in Bodenhamer et al. (2015). Qualitative methods are crucial to the analysis of oral histories and, as already
noted, corpus linguistics is an approach that we find especially promising to the study of narratives of the Holocaust,
specifically at scale—that is, from the dozens, to the hundreds, and, eventually, the thousands. Corpus linguistics and
natural language processing, a related branch of computer science, use computational methods to analyze words and
phrases in corpora, primarily in the context of linguistics research. However, these methods of text mining have been
applied to a broader set of research questions within digital humanities research, using analytical tools that include:

1. Concordance—displays a select word or phrase in the sentences in which it occurs.

3. Relative frequency—compares the frequency in the target corpus to the frequency in a larger reference corpus. For
example, comparing the use by female survivors of a term with the use by male survivors.
4. Collocation—returns the instances where two or more words occur in proximity to one another. For example,
street and fear.
5. Annotation—also known as tagging, annotation attaches information to words and phrases to enable semantic
analysis. For example, annotation can identify parts of speech and place names. Semantic tags identify words and
phrases as belonging to any of a number of preset conceptual categories, such as movement or family.

What attracts us in particular to corpus linguistics and natural language processing methods is that they enable a work-
ing at scale of hundreds, thousands, and tens of thousands of transcripts of oral history interviews with survivors. Pre-
liminary results are promising (Knowles, Jaskot, Cole, & Giordano, forthcoming 2019).
A GIS of place will not only necessitate a broadening of methods, but also require a rethinking of the collabo-
rative model that the two of us and the wider Holocaust Geographies Collaborative have employed in over 10
years of working together, in which we progressed (Giordano, Cole, & Knowles, 2014; Jaskot & Cole, 2014) from a
multidisciplinary to an interdisciplinary collaboration, and a transition to a transdisciplinary model, in which transdis-
ciplinarity is defined as the integration of multiple disciplines from the natural and social sciences that transcends
their traditional boundaries (revised from Choi & Pak, 2006). A multi-methods approach, collaborative research
across fields, and the integration of disciplinary perspectives are among the key elements of the digital and geospa-
tial humanities research model.
Spatial analysis and quantitative methods have a fundamental role to play in this research agenda, as they are the
interpretive tools to approach and investigate the abstract view of geography—as exemplified in Section 2 with cluster
analysis for Italy and network analysis for Budapest. This is a view that can be represented, measured, and understood
via quantitative data and methods; it is unambiguous and measurable with degrees of accuracy and precision; deals
with absolute (“miles,” “kilometers”) rather than relative (“near,” “far”) space; and is by definition designed to deal with
big data. It lacks, as already mentioned, the ability to deal effectively with the relative, the ambiguous, with multiple
interpretations of the same data piece (e.g., a word like “near”). This is by definition the victim’s view of the Holocaust:
narratives, emotions, ambiguities, social space, relative (e.g., “near”) rather than absolute space. This inability has been
noted already, of course, and several approaches—as discussed in the Introduction—have been proposed as a way to
overcome these limitations and as a step toward the integration of quantitative and qualitative methods. We believe
that steps toward this integration will require technical (e.g., new visualization methods) as well as conceptual and
methodological (model of reality, data model, analytical functionalities) solutions: for example, the ability to study
simultaneously social networks and spatial networks and how these move together and at scale—Budapest and
Buchenwald, the apartment and the barrack, and so on.
From a methodological point of view and in the specific context of this research, in order to define and investigate
the places of the Holocaust—and not only its spaces—we will need an integration of traditional spatial analytical meth-
ods with social networks analysis and corpus linguistics methods as tools to parse the testimonies of the survivors
 & Zalc, 2014). Building a GIS of place is beyond the scope of the topics
(for a Holocaust-specific example, see Merckle
discussed in this article, but we can start to sketch some of the challenges and solutions toward achieving the objec-
tive. In general, and specifically for the oral histories of the Holocaust we have become familiar with, testimonies tend
to lack the specificity of both time and place that traditional GIS requires. So, for example, for HGIS Budapest we can
tell with some certainty that on a certain date in 1944 a certain building—or an apartment in the building—was desig-
nated as a Jewish residence. This is not necessarily true for oral histories or testimonies. Thus, for example, humans
tend to narrate space in terms of relative as well as absolute space: “near” as well as, and perhaps more often than,
“10 km.” And they are likely to describe a certain place of hiding less in physical terms as they do by referring to the
social interactions they have with the people they are hiding with, the latter described in qualitative terms (“friends,”
“foes,” “he harmed me,” “she helped me”). Analogous considerations could be made for times, dates, events, which are
often ambiguous or lack precision and accuracy. We are currently working on a project to populate the buildings of
10 | GIORDANO AND COLE

1945 Budapest—only a few months after the city’s liberation in December 1944—with the approximately 75,000 sur-
viving Jews who lived there. Our hope is to then locate enough oral histories to start and build a GIS of place for Buda-
pest, and to tell its stories.
In our early exploration of appropriate methods for this complex data, it seems to us that relative spaces and rela-
tive times require a relative approach. In other words, a GIS of place needs to be by definition topological: it is the exis-
tence of a relation between people and place that takes center stage in a GIS of place. (Note the correspondence
between this view of GIS and the discussion of “integrated” history in the Introduction.) The object of study in a GIS of
place is therefore the relational and the topological. This is the deepest meaning of statements such as “the social is
spatial” and “social relationships are explicitly spatialized (i.e., they are inextricably and explicitly described as happening
in space).” As a way of spatializing social networks—or, if one prefers, adding the social dimension to spatial networks—
several authors have proposed methods to integrate social networks analysis and GIS: see, for example, Andris (2016),
especially the idea of an anthrospace and ways to measure it; and, in the context of the transmission of disease, Emch
et al. (2012) and Bian et al. (2012). We ourselves contributed to this literature with an article on the social and spatial
networks of the Budapest Ghetto (Giordano & Cole, 2011).
The challenge is how to map and represent the times and places of the testimonies, and to do it at scale (for all of
Budapest, for example): how do we build a GIS of place by interrogating not one or two but 2,000 testimonies and nar-
ratives? Scaling up is not at all an issue for traditional GIS—it is easy to georeference 3,000 addresses on a map of
Budapest, or to map the place of arrest for 6,000 Jewish victims of the Holocaust in Italy—but it is likely to be the
greatest conceptual and technical challenge for building a GIS of the places of the Holocaust. For a possible way of
thinking about how to deal with this challenge, we can think of the way that the census of population is created. Cen-
sus data are collected at the resolution of the individual and then “aggregated up” at different scales for different pur-
poses, and for zooming in or out of the study of a specific characteristic of the population. The aggregation is
accomplished by creating a “new” geography (in the USA, census blocks, blockgroups, etc.) and by “scaling up” from the
individual to the creation of “new” categories of individuals (women, Hispanics, households, etc.). Thus re-categorized,
the researcher asks the census dataset different questions depending on the scale of analysis, a scale that is geographi-
cal as well as thematic and temporal. Analogously, oral histories, testimonies, and narratives are also typically collected
individually, and they too can be “aggregated up” (this is what corpus linguistics can assist in achieving) (Shenker, 2015).
The job of the transdisciplinary collaborator—to use a term defined above—is to create aggregations that are both spe-
cific to the Holocaust and also rooted in a wider epistemology. For example, corpus linguistics can help us to categorize
the testimonies and also provide a base topological framework to start determining the spatial scales of a certain place
of the Holocaust (e.g., Budapest, Italy). The next step is which functionalities—analogous to cluster analysis or network
analysis in traditional GIS—a GIS of place will possess. Some of these functionalities will be borrowed, we predict, from
social network analysis, but studying and measuring social relations in place will require the amalgamation of GIScience,
social network analysis, corpus linguistics, and possibly other methods in ways that we are only beginning to explore.
To give just one very brief example that draws some of these tools and methods together, we return to the street
network of Budapest, which formed a key layer of the HGIS of the Budapest Ghetto that enabled us to undertake net-
work analysis. Working with corpus linguistic methods and oral history accounts suggests other ways of thinking about
the street, not just as a network but also as a place of encounters and emotions. Budapest was unusual in having both
a highly dispersed form of ghetto (spread across close to 2,000 apartment buildings that spanned the entire city) in the
summer and fall of 1944, as well as being a city where the shape and form of that ghetto was in constant flux across
1944, with a shift to two more distinct ghetto areas established in December 1944 (Cole, 2003). In both cases, the
street network played a particular role. During the summer and fall of 1944—as we noted in Section 2—the street was
the means of accessing key resources such as foodstuffs. In December 1944, the street was the site of Jews being
marched from one ghetto to another. What is striking from initial work with oral histories is how the street is remem-
bered and retold differently in these narratives across the second half of 1944. In particular, the collocation of the
word “street” and words associated with “fear” of violence against self and others is found in accounts of relocation in
December 1944. In short, the street network of the city, which formed a key part of our GIS of space, requires also a
GIORDANO AND COLE | 11

rendering within a GIS of place as occupied space where victims were brought into intimate and potentially deadly
contact with the armed perpetrators as they were marched to the Pest ghetto.

5 | CONCLUSIONS

Our long-term collaboration and the work of the Holocaust Geographies Collaborative—an international network of
historians and geographers interested in bringing geographical and GIScience perspectives to the study of the Holo-
caust—has shown that the tools and methods of GIS and GIScience have great potential for the digital and geospatial
humanities (Knowles, Cole, & Giordano, 2014). To more fully realize this potential, however, it seems to us we need to
work toward the realization of a GIS of place, intended as the integration of the traditional quantitative spatial analyti-
cal tools of GIS with qualitative data toward the full realization of a qualitative GIS. For this article, rather than outlining
the results of our work with a fully-fledged GIS of place, we have sought to articulate the necessity of new models and
what some key features of those new models might be. Our hope is that developing a mixed-methods GIS of place will
enable the kind of “integrated” and “relational” analysis that the complexity of place—whether historical or contempo-
rary—demands.

AC KNOW LEDG MENT S

We are grateful for the support of the National Science Foundation, the National Endowment for the Humanities,
and the Toni Schiff Memorial Fund for funding the research that underlies this article. We also wish to thank our
colleagues in the Holocaust Geographies Collaborative—Anne Kelly Knowles and Paul B. Jaskot—for their insights,
as well as Shelley Burleson, Seth Neuman, and Michele Tucci for the construction of the Historical GIS of
Budapest and Italy, and the Department of Geography at Texas State University and the Department of History
at the University of Bristol for additional support.

OR CI D

Alberto Giordano http://orcid.org/0000-0001-9427-9521

RE FE RE NCE S
Aitken, S. (2015). Quelling imperious urges: Deep emotional mappings and the ethnopoetics of space. In D. J.
Bodenhamer, J. Corrigan, & T. M. Harris (Eds.), Deep maps and spatial narratives (pp. 102–133). Bloomington, IN:
Indiana University Press.
Andris, C. (2016). Integrating social network data into GISystems. International Journal of Geographic Information Science,
30(10), 2009–2031.
Bailey Kellog, C., & Zhao, F. (2004). Qualitative spatial reasoning: Extracting and reasoning with spatial aggregates.
AI Magazine, 24(4), 47–60.
Bian, L., Huang, Y., Mao, L., Lim, E., Lee, G., Yang, Y., . . . Wilson, D. (2012). Modeling individual vulnerability to communi-
cable diseases: A framework and design. Annals of the Association of American Geographers, 102(5), 1016–1025.
Bodenhamer, D. J., Corrigan, J., & Harris, T. M. (Eds.). (2010). The spatial humanities: GIS and the future of humanities
scholarship. Bloomington, IN: Indiana University Press.
Bodenhamer, D. J., Corrigan, J., & Harris, T. M. (Eds.). (2015). Deep maps and spatial narratives. Bloomington, IN: Indiana
University Press.
Burleson, S., & Giordano, A. (2016). Spatiality of the stages of genocide: The Armenian case. Genocide Studies &
Prevention: An International Journal, 10(3), 39–58.
Choi, B. C., & Pak, A. W. (2006). Multidisciplinarity, interdisciplinarity and transdisciplinarity in health research, services,
education and policy: 1. Definitions, objectives, and evidence of effectiveness. Clinical & Investigative Medicine, 29,
351–364.
Cohn, A. G., & Hazarika, S. M. (2001). Qualitative spatial representation and reasoning: An overview. Fundamenta Informa-
ticae, 46, 1–29.
12 | GIORDANO AND COLE

Cohn, A. G., & Renz, J. (2007). Qualitative spatial representation and reasoning. In F. van Harmelen, V. Lifschitz, & B. Porter
(Eds.), Handbook of knowledge representation (pp. 551–596). Amsterdam, the Netherlands: Elsevier.
Cole, T. (2003). Holocaust city: The making of a Jewish ghetto. New York, NY: Routledge.
Cole, T. (2016). Holocaust landscapes. London, UK: Bloomsbury.
Cole, T., & Giordano, A. (2014a). Bringing the ghetto to the Jew: The shifting geography of the Budapest ghetto. In A.
Knowles, T. Cole, & A. Giordano (Eds.), Geographies of the Holocaust (pp. 121–157). Bloomington, IN: Indiana Univer-
sity Press.
Cole, T., & Giordano, A. (2014b). Rethinking segregation in the ghetto: Invisible walls and social networks in the dispersed
ghetto in Budapest, 1944. In H. Earl & K. Schleunes (Eds.), Lessons and legacies XI: Expanding perspectives on the Holo-
caust in a changing world (pp. 251–264). Evanston, IL: Northwestern University Press.
Cope, M. S., & Elwood, S. (2009). Qualitative GIS: A mixed methods approach. Thousand Oaks, CA: Sage.
Curry, M. R. (1998). Digital places: Living with geographic information technologies. London, UK: Routledge.
Dear, M., Ketchum, J., Luria, S., & Richardson, D. (Eds.). (2011). GeoHumanities: Art, history, text at the edge of place.
London, UK: Routledge.
Donaldson, C., Gregory, I., & Taylor, J. (2017). Implementing corpus analysis and GIS to examine historical accounts of
the English Lake District. In P. Bol (Ed.), Historical atlas: Its concepts and methodologies (pp. 152–172). Seoul, Korea:
Northeast Asian History Foundation.
Drucker, J. (2012). Humanistic theory and digital scholarship. In M. K. Gold (Ed.), Debates in the digital humanities
(pp. 85–95). Minneapolis, MN: University of Minnesota.
Egenhofer, M. J., & Mark, D. M. (1995). Naive geography. In A. U. Frank & W. Kuhn (Eds.), Spatial information theory: A
theoretical basis for GIS (Lecture Notes in Computer Science Vol. 988, pp. 1–15). Berlin, Germany: Springer.
Emch, M., Root, E. D., Giebultowicz, S., Ali, M., Perez-Heydrich, C., & Yunus, M. (2012). Integration of spatial and social
network analysis in disease transmission studies. Annals of the Association of American Geographers, 105(5),
1004–1015.
Frank, A. U., & Mark, D. M. (1991). Language issues for geographical information systems. In D. J. Maguire, M. F. Good-
child, & D. W. Rhind (Eds.), Geographical information systems: Principles and applications (pp. 147–163). London, UK:
Longman.
Friedländer, S. (1997). Nazi Germany and the Jews: The years of persecution, 1933–1939. New York, NY: Harper Collins.
Giordano, A., & Cole, T. (2011). On place and space: Calculating social and spatial networks in the Budapest Ghetto.
Transactions in GIS, 15(Suppl. 1), 143–170.
Giordano, A., Cole, T., & Knowles, A. (2014). Geographies of the Holocaust. In A. Knowles, T. Cole, & A. Giordano (Eds.),
Geographies of the Holocaust (pp. 1–17). Bloomington, IN: Indiana University Press.
Giordano, A., & Holian, A. (2014). Retracing the “Hunt for Jews”: A spatio-temporal analysis of arrests during the
Holocaust in Italy. In A. Knowles, T. Cole, & A. Giordano (Eds.), Geographies of the Holocaust (pp. 52–86). Bloomington,
IN: Indiana University Press.
Goodchild, M. F. (2004). GIScience, geography, form, and process. Annals of the Association of American Geographers, 94
(4), 709–714.
Gregory, I., Cooper, D., Hardie, A., & Rayson, P. (2015). Spatializing and analyzing digital texts: Corpora, GIS, and places.
In D. J. Bodenhamer, J. Corrigan, & T. M. Harris (Eds.), Deep maps and spatial narratives (pp. 150–178). Bloomington,
IN: Indiana University Press.
Gregory, I., & Donaldson, C. (2016). Geographical text analysis: Digital cartographies of Lake District literature. In D.
Cooper, C. Donaldson, & P. Murrieta-Fores (Eds.), Literary mapping in the digital age (pp. 67–78). London, UK:
Routledge.
Hilberg, R. (1961). The destruction of European Jews. Chicago, IL: Quadrangle Books.
Jaskot, P. (2015). Architecture of the Holocaust: The 2015 Joseph and Rebecca Meyerhoff Annual Lecture. Washington, DC:
U.S. Holocaust Memorial Museum.
Jaskot, P., & Cole, T. (2014). Afterword. In A. Knowles, T. Cole, & A. Giordano (Eds.), Geographies of the Holocaust
(pp. 227–232). Bloomington, IN: Indiana University Press.
Kawano, Y., Munaim, A., Goto, J., Shobugawa, Y., & Naito, M. (2016). Sensing space: Augmenting scientific data with
spatial ethnography. GeoHumanities, 2(2), 485–508.
Kim, J., Vasardani, M., & Winter, S. (2017). Similarity matching for integrating spatial information extracted from place
descriptions. International Journal of Geographic Information Science, 31(1), 56–80.
GIORDANO AND COLE | 13

€n, J. A. (2017). Qualitative spatial and temporal representation and reasoning. In D. Richardson,
Klippel, A., & Wallgru
N. Castree, M. F. Goodchild, A. Kobayashi, W. Liu, & R. A. Marston (Eds.), The international encyclopedia of geography
(pp. 5451–5458). Washington, DC: American Association of Geographers & Wiley Blackwell.
Knowles, A., Cole, T., & Giordano, A. (Eds.). (2014). Geographies of the Holocaust. Bloomington, IN: Indiana University Press.
Knowles, A., Jaskot, P., Cole, T., & Giordano, A. (forthcoming 2019). Mind the gap: Reading across the Holocaust testimo-
nial archive. In T. Cole & S. Gigliotti (Eds.), Lessons and legacies XIV: The Holocaust in the 21st century: Relevance and
challenges in the digital age. Evanston, IL: Northwestern University Press.
Knowles, A., Westerveld, L., & Strom, L. (2015). Inductive visualization: A humanistic alternative to GIS. GeoHumanities, 1
(2), 233–265.
Kwan, M.-P. (2002). Feminist visualization: Re-envisioning GIS as a method in feminist geographic research. Annals of the
Association of American Geographers, 92(4), 645–661.
Kwan, M.-P., & Ding, G. (2008). Geo-narrative: Extending geographic information systems for narrative analysis in
qualitative and mixed-methods research. Professional Geographer, 60(4), 443–465.
Lefebvre, H. (1991). The production of space. Malden, MA: Blackwell (In French, 1974).
Mennis, J., Mason, M. J., & Cao, Y. (2013). Qualitative GIS and the visualization of narrative activity space data.
International Journal of Geographical Information Science, 27(2), 267–291.
, P., & Zalc, C. (2014). Trajectories of the persecuted during the Second World War: Contribution to a microhis-
Merckle
tory of the Holocaust. In P. Blanchard, F. Bu €hlmann, & J.-A. Gauthier (Eds.), Advances in sequence analysis: Theory,
method, applications (pp. 171–190). New York, NY: Springer.
O’Sullivan, D. (2006). Geographical information science: Critical GIS. Progress in Human Geography, 30(6), 783–791.
Obermeyer, N. J. (1995). The hidden GIS technocracy. Cartography & Geographic Information Systems, 22(1), 78–83.
Pavlovskaya, M. (2017). Qualitative GIS. In D. Richardson, N. Castree, M. F. Goodchild, A. Kobayashi, W. Liu, & R. A.
Marston (Eds.), The international encyclopedia of geography (Vol. 10, pp. 5432–5442). Washington, DC: American Asso-
ciation of Geographers & Wiley Blackwell.
Picciotto Fargion, L. (1991/2002). Il libro della memoriaMemoria: Gli ebrei Ebrei deportati Deportati dall’Italia (1943–1945)
[The book of memory: Jews deported from Italy (1943–1945)]. Milan, Italy: Mursia.
Pickles, J. (1995). Ground truth: The social implications of geographic information systems. New York, NY: Guilford Press.
Rayson, P., Reinhold, A., Butler, J., Donaldson, C. E., Gregory, I., & Taylor, J. (2017). A deeply annotated testbed for geo-
graphical text analysis: The corpus of Lake District writing. In Proceedings of the First ACM SIGSPATIAL Workshop on
Geospatial Humanities (pp. 9–15). Redondo Beach, CA: ACM.
Salmoni, G. (2016). Buchenwald, una storia da scoprire [Buchenwald, a story to be discovered]. Genboa, Italy: Fratelli Frilli
Editore.
Schuurman, N. (2000). Trouble in the heartland: GIS and its critics in the 1990s. Progress in Human Geography, 24(4),
569–590.
Shenker, N. (2015). Reframing Holocaust testimony. Bloomington, IN: Indiana University Press.
Sheppard, E. (1995). GIS and society: Towards a research agenda. Cartography & Geographic Information Systems, 22(1), 5–16.
Smith, B., & Mark, D. M. (2001). Geographical categories: An ontological investigation. International Journal of Geographic
Information Science, 15(7), 591–612.
Song, G., Janowicz, K., Montello, D. R., Hu, Y., Yang, J.-A., McKenzie, G., . . . Yan, B. (2017). A data-synthesis-driven
method for detecting and extracting vague cognitive regions. International Journal of Geographic Information Science,
31(6), 1245–1271.
Thatcher, J., Bergmann, L., Ricker, B., Rose-Redwood, R., O’Sullivan, D., Barnes, T. J., . . . Young, J. C. (2016). Revisiting
critical GIS. Environment and Planning A, 48(5), 815–824.
Travis, C. (2015). Abstract machines: Humanities GIS. Redlands, CA: Esri Press.
Tuan, Y. (1977). Space and place. The perspective of experience. Minneapolis, MN: University of Minnesota Press.
Watts, P. R. (2010). Mapping narratives: The 1992 Los Angeles riots as a case study for narrative-based geovisualization.
Journal of Cultural Geography, 27(2), 203–227.

How to cite this article: Giordano A, Cole T. The limits of GIS: Towards a GIS of place. Transactions in GIS.
2018;00:1–13. https://doi.org/10.1111/tgis.12342