Conference PaperPDF Available

The Anatomy of Knowledge: Quantitative and Qualitative Analysis of the Evolution of Ideas in Space Syntax Conference Articles (1997-2017)

Abstract and Figures

Since its inception in the 1970s, space syntax has matured into a theory and a method comprising a set of recurring theoretical and analytical concepts, as well as new ones emerging through the years. How can we trace the evolution of the field through language? How can we analyse the development of ideas in space syntax research? What can we learn from this evolution about knowledge creation in this area? Recognising that language is central to the development of ideas in any field, this paper uses automated text-analyses, focusing more specifically on all papers published in the space syntax symposia proceedings from 1997 to 2017. The purpose is to trace the trajectory of ideas as they were elaborated, used and perhaps changed in the collective work of authors researching within this field in different parts of the world. Firstly, we identify concepts and technical terminology in the field through a combined quantitative and qualitative text analysis. Secondly, we statistically assess the use of these terms, revealing patterns and trends in the evolution of knowledge in space syntax. Thirdly, we compare patterns between established concepts and categories that stabilise over time with concepts emerging more recently. The results from our analysis of networks of concept relationships suggest that: (i) concepts and terms evolve in dependent trajectories; (ii) ideas have evolutionary developments, with some emerging and gaining growing attention, while others showing clear signs of stability, and others losing centrality over time, including networks of what can be termed as ‘canonical’ concepts. We have also identified (iii) an overall decline in the use of early space syntax concepts rooted in social theory and anthropology; (iv) a trend of decreasing conceptual novelty over time; (v) traces of increasing influence by other fields; and finally (vi) signs of a clear ‘technological turn’ in the field.
Content may be subject to copyright.
Proceedings of the 12th Space Syntax Symposium
137-1
THE ANATOMY OF KNOWLEDGE:
Quantitative and Qualitative Analysis of the Evolution of Ideas in Space Syntax
Conference Articles (1997-2017).
KIMON KRENZ1; SOPHIA PSARRA2; VINICIUS M. NETTO3
ABSTRACT
Since its inception in the 1970s, space syntax has matured into a theory and a method comprising a set
of recurring theoretical and analytical concepts, as well as new ones emerging through the years. How
can we trace the evolution of the field through language? How can we analyse the development of ideas
in space syntax research? What can we learn from this evolution about knowledge creation in this area?
Recognising that language is central to the development of ideas in any field, this paper uses automated
text-analyses, focusing more specifically on all papers published in the space syntax symposia
proceedings from 1997 to 2017. The purpose is to trace the trajectory of ideas as they were elaborated,
used and perhaps changed in the collective work of authors researching within this field in different
parts of the world. Firstly, we identify concepts and technical terminology in the field through a
combined quantitative and qualitative text analysis. Secondly, we statistically assess the use of these
terms, revealing patterns and trends in the evolution of knowledge in space syntax. Thirdly, we compare
patterns between established concepts and categories that stabilise over time with concepts emerging
more recently. The results from our analysis of networks of concept relationships suggest that: (i)
concepts and terms evolve in dependent trajectories; (ii) ideas have evolutionary developments, with
some emerging and gaining growing attention, while others showing clear signs of stability, and others
losing centrality over time, including networks of what can be termed as ‘canonical’ concepts. We have
also identified (iii) an overall decline in the use of early space syntax concepts rooted in social theory
and anthropology; (iv) a trend of decreasing conceptual novelty over time; (v) traces of increasing
influence by other fields; and finally (vi) signs of a cleartechnological turn’ in the field.
KEYWORDS
Space Syntax, Quantitative Text Analysis, Qualitative Text Analysis, Concept Identification, Concept
Trajectories
1 Kimon Krenz Space Syntax Laboratory, The Bartlett School of Architecture, University College London, 22 Gordon Street, London
WC1H 0QB, United Kingdom k.krenz@ucl.ac.uk
2 Sophia Psarra Space Syntax Laboratory, The Bartlett School of Architecture, University College London, 22 Gordon Street, London
WC1H 0QB, United Kingdom s.psarra@ucl.ac.uk
3 Vinicius M. Netto Universidade Federal Fluminense, Rua Passo da Pátria, 156, Red Beach Campus, Saint Domingos, Niterói, Rio de
Janeiro, 24210-240, Brazil vmnetto@id.uff.br
1
Proceedings of the 12th Space Syntax Symposium
1. INTRODUCTION
Since its inception in the 1970s, space syntax has matured into a theory and a method comprising a set
of recurring theoretical and analytical concepts and terminologies emerging through the years. Up to
the present, this research field has involved the production of over 1,000 conference papers texts that
register the collective contribution made by more than 1,000 researchers working within what came to
be a consolidated field of scientific inquiry into socio-spatial systems, buildings, cities, urban spaces
and even regions. Considering such an increasing pool of texts registering the emergence, consolidation
and continuing production of space syntax theory and methodological approach, can we now observe
the conceptual paths taken within the field? Can we trace the evolution of space syntax through the use
of the specific language developed in the field? What can such an analysis reveal of the development
of ideas in space syntax research? And what can we learn from this evolution about knowledge creation
in this field?
Philosophers of science like Thomas Kuhn (1962) made us aware that every scientific field or paradigm
encompasses their own sets of concepts, terminologies and techniques, which compose the specific
vocabulary and language used to approach and understand their object domain. Research directions and
outputs go hand in hand with the evolution of languages (discursive or otherwise) employed in the work
of researching and writing. Recognising that language is central to the development of ideas in any field,
this paper analyses the body of words that compose the language developed and used in Space Syntax.
For that, it uses automated and analogical text-analyses, focusing specifically on all papers published
in the space syntax symposia proceedings from 1997 to 2017 more precisely, 1,089 articles.
The purpose is to map the trajectory of ideas as they were developed, used and changed in the collective
work of authors researching in space syntax in different parts of the world. Space syntax symposia
papers are particularly interesting for the linguistic anatomy of the field. This is because these symposia
attract researchers engaging with ideas and terminology that are widely shared within the field.
Furthermore, conference publications are likely to bring fresh applications, attempts at further
developments, and new insights. Authors are more likely to test ideas in conferences in ways they hardly
could in a journal submission. As journals generally have high standards regarding robustness in
empirical samples and findings, conferences seem interesting platforms to test novelty. Moreover, space
syntax symposia can provide consistent temporal insights into the development of the field, due to their
regular two-year interval. In short, symposia papers seem to be a proper field of written registers to
prospect novelty, tentative uses and potential innovations in terminology, concepts and methods. Such
papers might reveal trends before authors become self-conscious about them.
The method deployed in this analysis involves the following steps. First, we identify concepts in the
field through a combined qualitative and quantitative text analysis that processes large amounts of texts
and identifies scientific concepts and terminologies. These terminologies encompass theoretical
concepts (such as ‘solidarity’ or ‘visibility’), methodological concepts (such as ‘integration’ or ‘choice’)
and technical concepts (such as ‘choice value’ or ‘angular analysis’). Second, we statistically analyse
the frequencies of the occurrence of these concepts and terms, and assess their evolution in time. Third,
we identify the changing networks of relationships created among these notions, and compare patterns
between established concepts that stabilise over time with concepts emerging more recently. Precisely,
we are constructing a network of concepts and terms through their contextual relationships and
investigate the structure and morphology of this network. This process reveals patterns and trends in
the evolution of knowledge in space syntax. Such comparisons might enable an understanding of the
relationship between what can be termed the ‘canonical’ (stable) structure of ideas with concepts
occurring less persistently in the body of papers.
We maintain that an automated text-analysis cannot replace a close reading of a text. What this method
offers is amplification and augmentation of careful reading and analysis. Although our research is
essentially inductive, i.e. a search for an understanding of the problem at hand driven by the empirical
data, as opposed to an investigation of empirical situations driven by a theoretical proposition, a series
of questions are motivating the project: What are the changes in substantive focus regarding the use of
theoretical and methodological concepts? What is the weight of theoretical, methodological and
technical work currently in the field? How are early concepts and the architectural, social and
anthropological framework (e.g. Hillier et al. 1976; Hillier and Hanson, 1984), which established space
syntax in the 1970s and 1980s holding in relation to new interests, technical innovations, analytical and
computational power, developed through the years?
2
Proceedings of the 12th Space Syntax Symposium
The analysis shows that concepts and terms are deeply relational: they can be understood through their
network structure of their relationships. In such a network concepts and terms appear in groups, and
find different positions and centralities in the topology of words used in the field. These centralities
change in time, and these distributions are quite revealing about the evolution of the field. The results
from analyses of concept frequencies, temporal trajectories and the network of concept relationships
found in these works suggest that: (i) concepts and terms evolve in dependent trajectories; (ii) ideas
have evolutionary developments, with some emerging and gaining growing attention, others showing
clear signs of stability, while others declining. This includes networks of ‘canonical’ concepts, which
may lose centrality over time. We have also identified (iii) an overall decline in the use of early concepts
rooted in social theory and anthropology, like ‘social solidarity’ and ‘encounter’; (iv) an apparent trend
of decreasing conceptual novelty over time; (v) traces of increasing influence by other fields, like
network science; and finally (v) strong signs of a ‘technological turn’, which describes a shift of the
field’s focus influenced by and towards technology i.e. the development of techniques, tools and
knowledge that is based on technological progress.
2. TEXT-ANALYSIS: METHODS
From the outset of the posed research questions: (a) how to understand the evolution of a scientific
field through its language’, (b) ‘how to trace the linguistic development of terminology and concepts’
and (c) ‘how to gain insights into the evolution of knowledge creation’, we have organised our approach
into three main steps, identifying concepts, tracing concepts over time, and identifying conceptual
systems, that are based on methods of textual data analysis.
2.1 LINGUISTIC IDENTIFICATION
A concept can be defined as an abstract representation or definition of an entity or phenomenon,
generally in the form of a generic idea which has been generalised from particular instances (Merriam-
Webster 2019; Saitta and Zucker 2013); it is through the formulation of concepts that we create
generalised theoretical understanding. The identification of concepts through quantitative methods is a
challenging endeavour. What particularly constitutes a concept depends highly on the theoretical
framework or field a concept is embedded into (Blumer 1931); this makes the formulation of a positive
definition, i.e. based on the existence of actual properties or components, difficult to be achieved. While
machines outperform humans in tasks such as processing large amounts of texts, they are still weak
with more complex tasks such as the extraction of meaning from texts. Extracting the generality of
meaning from individual words and sentences for a research field, however, is a fundamental component
for the identification of concepts. For this research, we propose a mixed approach of quantitative and
qualitative methods to identify theoretical and methodological concepts along with technical
terminology. For the quantitative part, a bag-of-wordsmodel (Harris 1954) is employed to establish a
list of words ranked according to their frequency across the database of conference articles.
The bag-of-words model is a simplified representation, where texts and documents are stripped from
word order and grammatical information and are instead combined into a ‘bag’ of all words maintaining
the number of a words occurrence. This model is commonly used in natural language processing and
information retrieval, and builds on the core assumption that a word’s frequency relates to its importance
within a text. Word frequency distributions have been the core interest in the field of statistical
linguistics for the past 85 years, grounded in linguist Georg Zipf’s empirical identification of an inherent
Pareto like distribution in natural languages. The so called Zipf’s law states that if a set of elements
for example, the words of a text are ordered by their frequency, the probability p of their occurrence
is inversely proportional to the position n within the order of rank: 𝑝𝑛 ~ $
% (Zipf 1932). This implies
that some words occur substantially more often than others, allowing the identification and
interpretation of their role within a language. In this research, word frequencies are computed for single
words (unigrams) and word combination of two words (bigrams) with the use of the R package
quanteda: for quantitative analysis of textual data (Benoit et al. 2018; R Development Core Team 2018).
The general aim is to identify concepts and terminologies that are used frequently rather than those used
only once.
The basis of this analysis forms a database of all papers published in the space syntax symposia
proceedings from 1997 to 20174. The database differentiates each paper by a unique ID, a combination
4 Short papers have not been included in the analysis, due to inconsistent information, quality and format.
3
Proceedings of the 12th Space Syntax Symposium
of the conference number, the sequential proceeding number, the first author, the university location
and country (e.g. 01_001_Hillier_London_GB, for the first space syntax paper by Hillier and Hanson
‘The Reasoning Art: or, The Need for an Analytical Theory of Architecture’ (1997)). All publications
are converted into a TXT document type and manually cleaned from page numbers, author and
affiliation information, as well as the list of references at the end of each paper. Moreover, the database
has been further cleaned through automated methods (i.e. regular expressions and quanteda based stop-
word removals). More specifically, we removed from the database words with little lexical meaning,
and those that primarily express grammatical relationships among other words (i.e. function words and
stop-words). Specifically, these words are: English stop-words (these refers to the most common words
in English language, e.g. ‘I’, you, to, etc.), conjunctive adverbs (e.g. accordingly, furthermore,
moreover, etc.), subordinating conjunctions (e.g. after, although, whenever, etc.), auxiliary verbs
(e.g. could, should, would, etc.), most common verbs (e.g. do, know, like, etc.), and most
common adverbs (e.g. accidentally, actually, afterwards, etc.), as well as a small number of
scientific jargon (e.g. et al, proceeding, fig, etc.). The result of this process will be referred to from
now on as corpus. A corpus can be subdivided into: documents, in our case conference papers; their
tokens, the collection of all words within a document; as well as types, the collection of unique words
within a document.
The number of publications in the corpus is 1089 with a total of 46601 unique selected words. Through
the ratio of types to tokens, i.e. token-type ratio (TTR) we can gain insights into the complexity of a
language or body of text (Ure 1971). TTR is the ordinary between the number of types (unique words)
and the number of tokens (total words) 5. The summary statistics of the cleaned corpus of conference
proceedings shows a decrease of the token-type ratio (TTR) from 0.076 to 0.044 during the period of
observation (Table 1). This means an increase in publications over time does not relate to an increase
in linguistic complexity and instead points to an increase in linguistic reproduction in the field.
Tab l e 1: Database summary statistics of the cleaned corpus
Year
1997
1999
2001
2003
2007
2009
2012
2013
2015
2017
Conference
01
02
03
04
06
07
08
09
10
11
Publications
39
47
67
79
110
115
91
111
152
177
Tokens
98823
80902
147542
199432
263682
268982
255864
286035
373929
435898
Types
7465
7606
10797
11841
13678
14234
13376
14165
16674
19282
TTR
0.076
0.094
0.073
0.059
0.052
0.053
0.052
0.050
0.045
0.044
The authors computed word frequencies for all types within the corpus and compiled a list of the 6,000
most frequent words. This list of 6,000 unigrams and bigrams was shown to three independent
researchers active in the field of space syntax with the request to mark/code all words that are or could
be characterised as concepts within the field. Figure 1 shows the frequency distribution of the 50 most
frequent unigrams and bigrams, with space being the most frequent unigram (occurring 42,867 times
and in 1084 papers) and space syntax the most frequent bigram (occurring 8,723 times and in 980
papers). For comparison, the 3,000th unigram is organizing(occurring 105 times and in 71 papers) and
the 3,000th bigram is drunken behaviour(occurring 31 times and in 4 papers). We acknowledge that
starting from the 6,000 most frequent unigrams and bigrams can form a limitation for the identification
of very new concepts, as their frequency is inevitably too low to be in the top 3,000 tokens. However,
this is not seen as problematic, as it is not the aim to identify all potential concepts, but rather to arrive
with a bottom-up list of concepts whose development is frequent enough to be traced over time.
5 TTR is calculated through the following formula, where V refers to the total number of types and N refers to the total number
of tokens: 𝑇𝑇𝑅 = ) *
+ .
4
Proceedings of the 12th Space Syntax Symposium
Figure 1: The 50 most frequent unigrams and bigrams across all conferences.
2.2 TRACING CONCEPTS OVER TIME
Based on the list of identified concepts by the three researchers, trajectories of these are traced over
time. This is done by computing frequencies for each of the concepts per conference year, providing a
quantification of their respective usage over time. Temporal frequency distributions can provide insights
into the development of concepts, as they give indicators on consistent, decreasing or increasing usage.
We differentiate between three fundamental stages: namely, the emergence (the coming into existence
of a concept over time), consistency (the state of sustained usage of a concept over time), and decline
of a concept over time. For instance, if a concept is not present in early conferences, but occurs at some
point and is increasingly used over time, it is classified as an emerging concept. If a concept is frequently
used in early years, but less so in later conferences, is categorised as a declining concept, and likewise
if there is no significant change in frequency, a concept is classified as a consistent concept. This
classification should not be seen as ordinal but rather as a continuous number of increasing or decreasing
degree. Furthermore, we standardize the computed frequency to allow comparisons of trajectories
across years and differing frequencies (henceforth ‘scaled word frequency’). Firstly, frequencies are
divided by the total number of tokens per conference year in order to be able to compare frequencies
between different years, and secondly, frequencies are standardized by subtracting the mean and
dividing by the standard deviation of the entire vector of words frequencies. Finally, the data trend for
each word is calculated and their slope compared across all concepts. Based on this comparison,
concepts are grouped into the proposed categories.
2.3 IDENTIFYING CONCEPTUAL SYSTEMS
In addition to the conceptstrajectory, it is of particular interest to identify the relationship among
different concepts. Do some concepts form conceptual clusters? Do related concepts exhibit co-
dependent trajectories, or do some concepts relate to each other in an inverse relationship where over
time one concept replaces another? In order to address these questions, we computed the co-occurrence
for each word in relation to all other words to investigate the relationship between different concepts.
This is done by counting how often a concept occurs in each paper for all concepts and the entire corpus;
the resulting vectors of occurrences are then correlated with each other and form a word co-occurrence
matrix. This matrix can be visualized as a network of relationships. Within this network, edges are
undirected and weighted by the respective correlation value. Formal descriptions of network properties
and morphology provide insights into how and to which extent concepts relate to each other.
10000
20000
30000
40000
datum
segment
main
process
place
research
part
scale
type
global
location
new
way
axial
form
measure
activity
design
point
time
person
result
centre
configuration
level
within
number
house
model
syntax
structure
map
pattern
local
system
social
network
different
line
movement
building
value
integration
street
analysis
urban
area
city
spatial
space
token freqeuncy
● ●
2500
5000
7500
social_interaction
public_transport
integrated_space
natural_movement
urban_environment
syntax_theory
syntactic_analysis
urban_structure
step_depth
town_centre
visual_field
large_scale
road_network
residential_area
spatial_pattern
spatial_property
spatial_network
ring_road
spatial_system
living_room
ground_floor
urban_fabric
visibility_graph
socio_economic
urban_system
spatial_layout
urban_design
movement_pattern
street_segment
urban_grid
spatial_analysis
syntax_analysis
city_centre
urban_form
convex_space
co_presence
urban_area
urban_space
integration_core
local_integration
pedestrian_movement
spatial_structure
public_space
street_network
global_integration
axial_map
integration_value
axial_line
spatial_configuration
space_syntax
token freqeuncy
5
Proceedings of the 12th Space Syntax Symposium
3.0 RESULTS
3.1 CONCEPTS IN THE FIELD OF SPACE SYNTAX
Based on the previously introduced list of most frequent words the three researchers independently
identified 816 terms in total, of which 205 are unigrams and 611 are bigrams. For reasons of intercoder
reliability, i.e. the extent to which two or more independent researchers/coders agree on the coding, we
subsequently select those terms that are identified by at least two researchers. Only 287 (222 bigrams
and 65 unigrams) of these terms are classified as conceptsand ‘technical terminology’ by at least two
researchers. This means that three-fourths of the most frequent concepts in the field of space syntax are
predominantly bigrams, i.e. two-word combinations. These bigram concepts are for example words
such as (in unrelated order) foreground network’, ‘generic functionor isovist occlusivity. Examples
of identified unigrams are words such as interface’, ‘permeability, or choice. Table 2 shows the 20
most frequent of these 287 identified terms as well as the number of papers each occurs in.
Tab l e 2: 20 most frequent uni- and bigrams identified by researchers in the field, among a total of 287 terms.
#
unigram
frequency
No of
papers
bigram
frequency
No of
papers
1
integration
15401
933
space_syntax
8723
980
2
local
10147
856
spatial_configuration
3559
661
3
syntax
9510
1001
axial_line
2724
434
4
configuration
8018
923
integration_value
2619
528
5
axial
7206
668
axial_map
2452
455
6
global
6690
747
global_integration
1863
349
7
choice
4926
628
street_network
1852
319
8
accessibility
4531
607
spatial_structure
1598
458
9
depth
4470
596
local_integration
1312
258
10
core
3882
587
integration_core
1207
227
11
visibility
3665
418
co_presence
1040
215
12
path
3511
530
convex_space
1039
217
13
syntactic
3275
537
urban_grid
962
249
14
connectivity
3224
518
street_segment
881
184
15
node
3147
394
movement_pattern
803
267
16
isovist
2775
216
visibility_graph
674
158
17
centrality
2628
338
visual_field
532
158
18
visitor
2291
328
step_depth
530
100
19
configurational
2279
495
syntactic_analysis
524
219
20
angular
2169
265
natural_movement
494
191
These 287 terms can be further differentiated into categories according to their roles in the space syntax
language. They include: theoretical concepts, not just words representing entities but interpretations of
a property or a phenomenon through discursive means, rendering that property or phenomenon
reasonably knowable without particular dependence on methodological descriptions (e.g.
‘visibility’); methodological concepts, relying on analytic procedures developed for investigating
empirical problems, including representation and calculation, rendering a property or phenomenon
representable and quantifiable, say, through geometrical or mathematical descriptions (e.g. isovist,
isovist integration’); and technical concepts, including words expressing methodological procedures
or components, which on their own do not qualify as methodological concepts (e.g. ‘nach value’, a
standardised value of choice). The above classification is instrumental for the purposes of our analysis.
6
Proceedings of the 12th Space Syntax Symposium
Relationships internal to these categories should be subject for further study, as they are likely to provide
more precise information on the evolution of specific classes of ideas and their role in the overall
evolution of the field. At this stage, however, we shall focus on the entire system of connections among
words in use in space syntax.
3.1 CONCEPT TRAJECTORIES
The established series of keywords has been used to perform a temporal analysis of word frequencies.
Concepts are subsequently grouped according to their commonalities in terms of slope value in order to
classify a concept into identified categories (emergence, decline and consistency). The ten most
increasing, decreasing and consistent concepts are plotted in Figure 2 and Figure 3 respectively. The
share of increasing concepts (trend line slope > 0.10) is with 80 of 287, twice as big as the number of
decreasing concepts with 27 (trend line slope < -0.10), while the majority shows rather steady
developments across the observed time period with 180.
The overall trends of each of the ten concepts with most increasing frequencies are highly comparable,
with relative variation in their curve development (Figure 2). Differently to these common
developments, the time of emergence of concepts is one of the key differences. However, while some
terms (e.g. nach valueor angular distance) exhibit steadily increasing usage over time, others (e.g.
betweennessor background network) increase showing higher variation in their curve development.
The ten concepts with most decreasing frequencies exhibit much higher variation in their curve
compared to the ten most increasing concepts (Figure 3, top). Most concepts have been used highly in
early conferences, after which their use declined rapidly within the first 5 conferences. Concepts such
as space occupancy and global movement for example where highly used in the first conference
(1997), declined shortly after and almost disappeared entirely. Others, such as axial graph, followed
the same trajectory, but experienced a sharp increase in 2007, just to disappear shortly after. Overall,
one can observe that there is a series of concepts that have significantly lost importance over time.
Trajectories of decline are not always steady processes and short temporary recurrences are not rare,
yet their overall decline is clearly visible.
A view on the ten concepts with slope values < 0.1 and > -0.1 unveils a third kind of trajectory, namely
concepts that feature no substantial increase or decrease over time (Figure 3, bottom). Such concepts
have mostly been present in all conference years and can be identified as the concepts definingspace
syntax canon. While these observations provide insights into the general trajectories of concepts in the
field of space syntax, little has been said about moments of concept emergence yet.
Figure 2: Concept emergence: The 10 most increasing concepts and terms. Scaled word frequencies per conference year
(1997 to 2017) with superimposed trend lines. slope value?
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
1997 1999 2001 2003 2005 2007 2009 2012 2013 2015 2017
year
scaled word frequency
concepts
angular distance
angular segment
background network
betweenness
betweenness centrality
choice value
closeness centrality
movement potential
nach value
segment map
7
Proceedings of the 12th Space Syntax Symposium
Figure 3: (Top) Concept decline: The 10 most decreasing concepts. Scaled word frequencies per conference year (1997 to
2017) with superimposed trend lines. (Bottom) Concept consistency: The 10 most consistent concepts. Scaled word
frequencies per conference year (1997 to 2017) with superimposed trend lines.
By focusing on years in which concepts occur for the very first time, we can identify those occurrences
of high linguistic and conceptual novelty as well as those lacking such. We have counted the occurrence
of new concepts per conference year (Table 3). By definition, the first symposium in 1997 was the one
with the highest number of ‘newly’ introduced concepts. Due to the difficulty in determining whether
all these concepts were truly ‘new’ at the time, we instead concentrated on the 2nd11th symposia. Here,
a clear trend of decreasing conceptual novelty is apparent. In 1999, 23 new concepts where introduced,
yet five conferences later in 2009 only three new concepts where introduced. The conferences in the
years 1999, 2001 and 2005 had the highest number of newly introduced concepts. Such a shift in
conceptual production might be related to an increasing endeavour to investigate the application of
concepts in research. At the same time, these results need to be considered with care, as concepts that
have very recently emerged, such as ‘avoidance’ (Koch 2015) might not have been utilised often enough
to feature in the initial list of 6,000 most frequent words. New concepts and terminologies introduced
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
1997 1999 2001 2003 2005 2007 2009 2012 2013 2015 2017
year
scaled word frequency
concepts
axial graph
geometrical
global movement
grid condition
integrating
multiplier effect
rra value
space occupancy
symmetry
virtual community
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
1997 1999 2001 2003 2005 2007 2009 2012 2013 2015 2017
year
scaled word frequency
concepts
co presence
complex system
configuration
emergence
node
permeability
topological
visitor
visual connection
visual depth
8
Proceedings of the 12th Space Syntax Symposium
especially in more recent years might not have found enough frequency to feature above the threshold
considered in our filter.
Tab l e 3: Number of newly emerged concepts per conference year.
Concepts
191
23
28
10
23
6
3
0
3
0
0
Year
1997
1999
2001
2003
2005
2007
2009
2012
2013
2015
2017
A closer look at the moment of emergence and development of newly introduced concepts can shed
light on this situation. Figure 4 and Figure 5 show plots of scaled word frequencies (see 2.2 for
definition) of all words that have been newly introduced in the conference years of 2003 and 2007, and
exhibit conceptual emergence. In 2003, a series of 10 concepts was introduced with different relative
frequencies. Here, not only the relative frequency at the moment of introduction differs substantially,
but also its development in the following years. Whereas for example, the use of the concept choice
measure was relatively low in the year of its introduction, it became consistently higher over the
following years, with little variation in its development. In contrast, the concept of visual depth, shows
a ‘M’-shaped development. First this concept was highly used in the year of its introduction, then not
used at all in 2005, followed by high usage in 2007 and 2009; finally, it disappeared entirely after 2009.
Such a spark of usage in one year, followed by decline also characterises concepts that emerged in 2007
(Figure 5). Here the concept of visibility costhas been introduced for the first time; its usage decreased
significantly in 2009 and was never used since. An example of concepts whose temporal development
indicates a related usage in text areforeground networkand background network; the development
of their curve shows that they are co-dependent or that there is a interrelationship between the two
concepts.
Figure 4: The 10 concepts that emerged in 2003. Scaled word frequencies per conference year (1997 to 2017).
● ● ●
● ● ●
● ●
● ● ●
● ● ●
● ● ●
● ●
● ● ●
● ● ●
● ● ●
● ● ●
● ●
● ● ●
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
1997 1999 2001 2003 2005 2007 2009 2012 2013 2015 2017
year
scaled word frequency
concepts
choice measure
dual graph
metric radius
network configuration
residential activity
segment model
synergy value
vga analysis
visual depth
visual graph
9
Proceedings of the 12th Space Syntax Symposium
Figure 5: The 10 concepts that emerged in 2007. Scaled word frequencies per conference year (1997 to 2017).
Investigating trajectories of concept usage over time has provided insights into the wide-spread
significance of some concepts of concepts; the moment of conceptual emergence; those concepts that
show identical or similar trajectories. However, this investigation does not capture the actual
relationships between two or more concepts. We can further explore these relations through network
analysis.
3.3 CONCEPTUAL SYSTEMS
To understand how concepts and terms are related to each other, we construct a network system of
related concepts. This network is based on significant correlations between the co-occurrences of
concepts, and allows the visualisation of the conceptual system (a system of relations among words
defining properties or phenomena) of the space syntax field. This is done by counting the occurrence of
each of the 287 concepts for the entire corpus (1089 papers). The result is a co-occurrence matrix (287
x 1089) where each concept is represented by a vector of 1089 values counting the occurrence within
each paper. These vectors are correlated with each other in all possible combinations (i.e. each concept
against every other concept) and the resulting correlation matrix forms the basis of the network creation.
Only concepts that feature a significant p-value (p <0.05) and a correlation coefficient above a threshold
of R2 >0.1 are then connected to each other. Figure 6 shows the result of this process: a network system
of all identified space syntax terms and concepts. The network edges are weighted according to their
correlation coefficient: the thicker the edge connection, the stronger the network relationship between
two concepts. To provide some examples of these relationships from Figure 6, strongly correlating
conceptual pairs are:movement economy––multiplier effect, description retrieval––spatial law’,
and social solidarity––transpatial’. The network node size is also weighted according to the network
degree, or connectivity of a node; this gives an indication of how many other concepts a particular
concept is related to (i.e. feature a correlation of R2>0.1); this is comparable to a concept’s connectivity,
to make use of space syntax terminology. Moreover, each node is coloured according to the computed
slope value of its trajectory trend line. Blue colours indicate declining usage over time, while red ones
highlight those concepts that are increasingly used. With the help of this system, one can not only
understand the relationship between two words but also potentially trace relationships through the
networksshortest path (where distance cost is the inverse correlation coefficient) between non-adjacent
concepts, pointing to potential establishment of theoretical connections. Examples of such a path are
from spatial practiceto affordance’;
spatial practice––social solidarity––affordance’,
or from the concept angular betweennessto convex partitioning’;
● ● ● ● ●
● ● ● ● ●
● ● ● ● ●
● ● ● ● ●
● ● ● ● ●
● ● ● ● ●
● ● ● ●
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
1997 1999 2001 2003 2005 2007 2009 2012 2013 2015 2017
year
scaled word frequency
concepts
background network
choice radius
foreground network
foreground structure
segment integration
visibility cost
10
Proceedings of the 12th Space Syntax Symposium
angular_betweenness––visibility cost––visual information––convex partitioning’,
or from the concept non discursiveto spatial order’;
non discursive––cognition––description retrieval––morphic language––spatial order.
The conceptual system also offers relevant insights into clusters of concepts. Clusters of concepts
feature substantially higher number of connections, as well as stronger correlation between concepts
within the cluster than to concepts outside of the cluster. We highlight four fundamental clusters in the
network. These four clusters are not only distinctive in their network relationships, but also in their
temporal development. Cluster 1, a dense network of predominantly emerging concepts; cluster 2 a
dense network of predominantly declining concepts; cluster 3 a sparser network with heterogeneous
usages of concepts, and cluster 4 a less densely connected network of declining theoretical concepts.
More significantly, the first two clusters are densely related to each other indicating the replacement of
one category of methodological concepts by another while also maintaining their relationship with each
other. In contrast, clusters 3 and 4 are more isolated. Therefore, there is intensification of a system of
emerging methodological concepts and one system of declining concepts of similar nature, as opposed
to a number of theoretical concepts that have sparser relationships with one another. This indicates a
lack of investment in tight interconnections among theoretical and methodological concepts or a
severance between analysis and theory, hence a decline of theoretical innovation.
Figure 6: Network of concept relationships. Edge thickness indicates the strength of relationship based on the correlation
coefficient, node size indicates the connectivity of concepts based on the network degree, while blue and red colours
highlight decreasing and increasing usage over time. Dashed lines highlight distinctive conceptual sub-clusters.
A closer look at the clusters shows the nature of these conceptual subsystems, which are distinctive
networks within the conceptual network. Figure 7, shows the network of cluster 1. The concept with the
highest degree is angular, followed by betweenness centralityand angular choice. All concepts are
11
Proceedings of the 12th Space Syntax Symposium
of rather methodological nature, with an emphasis on spatial network terminologies such as angularity,
segments, centrality, shortest path, and choice/betweenness. Since all of the concepts exhibit an
increasing trend, this cluster appears as one of the most important research directions of the field of
space syntax. We can also identify further sub-clusters within this conceptual network; betweenness
centrality, closeness centrality, network centralityform a distinctive sub network, as well as
technical terms like nach value, nain value, and normalized angular, which indicate specific sets
of words that are simultaneously used by authors employing any of these concepts.
Figure 7: Network of concept relationships: cluster 1. Edge thickness indicates the strength of relationship based on the
correlation coefficient, node size indicates the connectivity of concepts based on the network degree, while blue and red
colours highlight decreasing and increasing usage over time.
The second conceptual subsystem (cluster 2), features a different network morphology compared to
cluster 1 (Figure 8). All concepts exhibit declining trajectories and point to a strand of theoretical and
methodological terminology that loses its importance in the field. The concepts with the highest
connectivity in cluster 2 are integration, axialand global, to which two tree-like subgroups connect.
The first subgroup, connected to integration, defines the work on segregation with concepts such as
social segregationand ‘spatial segregation; the second subgroup, connected to global, is formed of
concepts such as intelligibilityand synergy’. Concepts that feature word pairs based on thelocalto
globalrelationship show a clear declining tendency. This declining tendency might indicate an earlier
argued need for new more complex differentiated conceptualisations of network relationships that go
beyond the dichotomy of ‘local’ and ’global (Krenz 2017). Overall, cluster 2 has more theoretical
concepts related to properties of a configurational nature such as ‘global’ or cognition such as
‘intelligibility’ as opposed to cluster 1, which features more methodological concepts. Furthermore,
cluster 2 seems to relate to the earlier methodological work in space syntax, when axial line maps
formed the basis of analysis. Compared to cluster 1, this early methodological work has declined over
12
Proceedings of the 12th Space Syntax Symposium
time, as the introduction of angular analysis (Turner 2000) and the subsequent method for angular
segment analysis breaking axial line into segments resulted in more and more studies employing
segment maps. This seems to coincide with a growing usage of concepts from network theory.
Figure 8: Network of concept relationships: cluster 2. Edge thickness indicates the strength of relationship based on the
correlation coefficient, node size indicates the connectivity of concepts based on the network degree, while blue and red
colours highlight decreasing and increasing usage over time.
The third conceptual subsystem (cluster 3) is not characterized by a clear trend in terms of the usage of
its concepts (Figure 9). Instead, there are seven increasing and seven decreasing concepts among ten
consistent ones. The concepts with the highest connectivity are isovist, visibility graphand visual
integration. These concepts are mostly used in the analysis of buildings and small-scale spaces and
point to a particular strand of research, which features a more heterogeneous development compared
to 01 and 02,
13
Proceedings of the 12th Space Syntax Symposium
Figure 9: Network of concept relationships: cluster 3. Edge thickness indicates the strength of relationship based on the
correlation coefficient, node size indicates the connectivity of concepts based on the network degree, while blue and red
colours highlight decreasing and increasing usage over time.
Finally, a fourth cluster shows the relative decline in the use of early theoretical concepts (Figure 10).
The network analysis of space syntax concepts shows that a constellation of methodological concepts
and technical terms has not only come to be proportionally dominant over more discursive and
theoretical terms, but sociological concepts and concepts related to the field of cognition studies like
‘virtual community’, ‘solidarity’, ‘encounter’ and ‘description retrieval’ have declined in usage.
14
Proceedings of the 12th Space Syntax Symposium
Figure 10: Network of concept relationships: cluster 4. Edge thickness indicates the strength of relationship based on the
correlation coefficient, node size indicates the connectivity of concepts based on the network degree, while blue and red
colours highlight decreasing and increasing usage over time.
4. CONCLUSIONS
Our analyses showed a number of trends within the space syntax field. As the field evolves, new
concepts emerge (e.g. ‘metric radius’, ‘topological radius’, ‘angular integration’), suggesting a search
for analytical advancement and terminological novelty (Appendix, Table 4). There is a clear split
between the uses of early theoretical and methodological spatial concepts (‘integration’, ‘axial line’)
and newer methodological terminology (‘choice’, ‘segment’) based on the decrease of the former and
the increase of the latter.
Previously defined primarily by axial analysis, increased computational power, new computational
techniques and the introduction of angular segment analysis allowed the field to connect to network
science and approaches using network analysis. This trend is indicated by the increasing use of network
concepts, like betweenness centrality, along with the shift from axial analysis to segment analysis,
with the growing exploration of GIS tools and road-centred maps (OpenStreetMap data), which can be
downloaded at no charge and are technically comparable to segment maps. Network-related terms are
replacing morphological terms. Accordingly, early terms from the Social Logic of Space (Hillier and
Hanson 1984) such as ‘relative asymmetry’ and ‘axial map’ are consistently less used since the year
1997 (Appendix, Table 5). Similarly, we identify a decline in topological analysis, i.e. an apparent shift
from step-based distance to metric or angular analysis where distance is based on angles independent
from topological steps. This is evident in the less frequent use of the word ‘topology’ (with the exception
of the terms ‘topological accessibility’, ‘topological choice’ and topological radius’). Angular analysis
emerges as a central methodological basis, along with the concepts of ‘closeness’ and betweenness
centrality’. Concepts like ‘occlusivity’ (Benedikt 1979) and ‘visibility analysis’ form a cluster of their
own, as VGA and Isovist analysis becomes more commonly used due to increasing computational
power (Benedikt and Mcelhinney 2019); this trend might also relate to the growing role of
computational methods and analytics in Building Information Modelling. Here we have examples of
the influence of technology and technical means shaping language use and research focus, thereby
shaping the field itself. Based on these results we can suggest that there is a clear trend to use
methodology, technique, technical development and analytical application as the main drive in the field,
advancing knowledge through the analytical side of space syntax, but paying less attention to knowledge
development through theoretical exploration.
15
Proceedings of the 12th Space Syntax Symposium
As the field opens up to network approaches, there is a growing recognition of theoretical parallels of
space syntax measures and earlier measures such as betweenness centrality and closeness centrality
(Freeman 1977; Sabidussi 1966). These standard network concepts are now explicitly acknowledged
by space syntax researchers, suggesting a growing openness towards network analysis in the field. This
also seems to be the case with the recent borrowing of concepts in space syntax conference papers, such
as ‘affordance’, ‘agent-based’, ‘entropy’ and ‘spatial practice’ (Appendix, Table 6). The notion of
‘affordance’, first introduced in 2001 and continuously rising since 2007, is a concept able to bridge
space, cognition and behaviour. After an initial peak in 2003, the concept ‘Agent-basedis consistently
increasing since 2012 and might be related to Agent-Based-Models (ABM), and trends in other fields
that use simulation of behaviour in an environment (Wallentin 2017). The higher frequency of the term
might also be associated with the use of the software tool DepthmapX. Finally, the concept of ‘entropy’
is a term that today cuts across many disciplines, from information theory and physics to biology, social
theory and urban studies (Gleick, 2011; Hidalgo, 2015; Davies, 2019; Bailey, 1990; Batty, 2014). Its
use within the space syntax field might be related to those other fields, ranging from assessing diversity
in the environment to describing its levels of order.
Previous critiques of space syntax raised concerns about the apparent isolation of the field and risks of
a self-referential terminology (Westin 2014, Netto 2016). Assessing these observations requires detailed
qualitative analysis of the conference papers utilizing these terms, which is a next step in our research.
Nevertheless, the increasing frequency of these concepts suggests the possibility of increasing influence
by other fields. On theoretical grounds, there are relatively few new concepts in the field. Concepts
related to the notion of morphogenesis like ‘generative process’, spatial law’, as well as ‘morphic
language’ and ‘description retrieval’ are declining after peaks in 2001 and 2003 respectively. In contrast,
concepts related to cognition (’spatial cognition’, ‘way-finding’, ’visual information’, ‘visibility’,
‘visual connectionandvisual integration’) are stable, while concepts related to ‘convexity’ are
declining in use possibly indicating decline of building studies in the area (Appendix, Table 7). One
exception is the increased use of the concept ‘visual control’. Importantly, we can observe a telling
decline in use of classic concepts rooted in social theory and anthropology, like ‘virtual community’,
‘solidarity’, ‘encounter’, ‘social structure’, and ‘segregated space’. Concepts related to broad definitions
about movement (‘movement economy’, ‘natural movement’, ‘movement pattern’) are also losing
importance. The increasing use of concepts such as ‘human behaviour’ and especially ‘spatial practice’,
which seems to be reminiscent of Lefebvre’s and De Certeau’s works, suggests signs of sociological
and behavioural explorations in alternative traditions to the Durkheimian framework proposed in Hillier
and Hanson (1984).
In a way, it is expected that a well-defined set of theoretical concepts guide a research field (Kuhn,
1962), including new methodological developments, but the decrease in usage of such concepts coupled
with the increase in the use of methodological and technical terms suggests that a scientific field is
developing along more methodological and technical directions. Of course, the extent that this is the
case must be subject to further scrutiny. On one hand, the apparent decline in innovation of theoretical
concepts since 2005 (see Table 3 on the number of newly emerged concepts per conference year) may
be seen as a function of the growing stability of a theory as a paradigm, in Kuhn’s sense of a theory
established around a set of concepts and methodological rules that ensure the coherence of its
applications and further theoretical and methodological developments. On the other hand, it might be
related to a growing conservatism apparent in the recursive use of its own conceptual terminology and
its reproduction as ‘normal science’, to use Kuhn’s words. Alternatively, growing competition from
other fields such as network science might have played a role, inspiring space syntax researchers
towards technologically driven explorations. Network science offers significant developments in the
area of environmental research.
However, from the outset, space syntax provided a new way of describing the socio-spatial dimensions
of buildings and cities not simply by representing and measuring spatial relationships through graph
analysis, but also by providing a theory of space, one which ‘should account for how and why different
societies generate different spatial patterns rather than interpreting a variable (different societies or
social patterns) by a constant (one particular instance of behaviour or observable phenomenon).
According to Hillier and Hanson, as opposed to analysing space, analysing behaviour and looking at
their relationship, a theory of space considers that ‘society already pervades those patterns that need to
be analysed while space carries social determination in its very form as object (1984 p. 8). A set of
‘postulates’ (ibid. p. 95-97) linking different categories and measures of space with social categories in
the Social Logic of Space established what the authors described as an ‘interpretive framework’, a
layered and structured set of relationships between spatial morphology, spatial measures and socio-
16
Proceedings of the 12th Space Syntax Symposium
spatial interpretation (ibid.). Therefore, the strength of space syntax has historically lied not simply in
its quantifiable and technological aspects, but also in the systemic interrelationship of conceptual and
methodological ideas, spatial and social concepts. The rising emphasis on network ideas and
methodology at the dispense of theoretical concepts indicates that the link between the theoretical and
methodological side of space syntax has been severed defining a technological turn in the field. The
larger implication is that when research aiming to explain how space is created for social purposes,
whether by design or accumulatively, is solely driven by technological determinism, it makes questions
such as what a good space consists of nearly impossible, for the simple reason that the answer has the
empty force of truism. It is already built into the supposition and hence, is self-evident: technology.
Of course, we must take into account that even new concepts would have to find a high enough
frequency to appear among the 6,000 most used words in space syntax, and be included in our analysis.
Nevertheless, confirming earlier observations (Griffiths and Netto 2015; Netto 2016), it is reasonable
to assert that there is a clear technological turn’ in the field, one in which conceptual ideas increasingly
have a ‘ghosted’ rather than active presence (Psarra, 2009). The next steps of research on the evolution
of concepts in space syntax will look into subsets of words, namely the specific networks of theoretical
concepts, methodological concepts, and technical terms, in order to see their relationships in more detail.
We also expect to assess:
§ The emergence of ideas in space syntax in relation to global academic discussions on
comparable topics in other fields, such as approaches to street networks (Marshall et al 2018),
spatial interaction (Batty et al 2014) and so on.
§ The spatial distribution of usage and creation of terms and concepts in relation to geographic
contingencies and potential regional intellectual clusters.
§ The trajectories of specific terms: how they have emerged or declined, through qualitative
analysis of sources.
GLOSSARY
ABM: Agent-Based Modelling
SLS: The Social Logic of Space
TTR: Types-Token Ratio
VGA: Visibility Graph Analysis
REFERENCES
Bailey, K.D. (1990). Social Entropy Theory. New York: State University of New York Press.
Batty, M., Morphet, R., Masucci, P. and Stanilov, K. (2014). Entropy, complexity, and spatial
information, Journal of Geographical Systems, 16/4: 363385.
Benedikt, M. L. (1979). To take hold of space: Isovists and Isovist Fields. Environment and Planning
B, 6: 4765.
Benedikt, M. L., and Mcelhinney, S. (2019 preprint). Isovists and the Metrics of Architectural Space.
Ficca J. and Kulper A. (eds) Proceedings 107th ACSA Annual Meeting, pp. 110. Pittsburgh, PA,
United States: ACSA Press.
Benoit, K., Watanabe, K., Wang, H., Nulty, P., Obeng, A., Müller, S., and Matsuo, A. (2018). quanteda:
An R package for the quantitative analysis of textual data. Journal of Open Source Software, 3/30:
774. DOI: 10.21105/joss.00774
Blumer, H. (1931). Science Without Concepts. American Journal of Sociology, 36/4: 51533.
Merriam-Webster. (2019). 'concept'. Merriam-Webster.com. Retrieved March 9, 2019, from
<https://www.merriam-webster.com/dictionary/concept>
Davies, P. (2019). The Demon in the Machine: How Hidden Webs of Information Are Finally Solving
the Mystery of Life. London: Penguin Books
17
Proceedings of the 12th Space Syntax Symposium
Freeman, L. C. (1977). A Set of Measures of Centrality Based on Betweenness. Sociometry. DOI:
10.2307/3033543
Gleick, J. (2011). The Information: A History, A Theory, A Flood. New York: Vintage.
Griffiths, S., and Netto, V. M. (2015). Open Syntaxes: Towards New Engagements with Social Sciences
and Humanities. The Journal of Space Syntax, 6/1: iv.
Harris, Z. S. (1954). Distributional Structure. WORD, 10/23: 14662. DOI: 10.1080/00437956.
1954.11659520
Hidalgo, C. (2015). Why Information Grows: The Evolution of Order, from Atoms to Economies. New
York: Basic Books.
Hillier, B., and Hanson, J. (1984). The Social Logic of Space. Cambridge: Cambridge University Press.
——. (1997). The Reasoning Art: or, The Need for an Analytical Theory of Architecture. Space Syntax
First International Symposium, 6.
Hillier, B., Leaman, A., Stansall, P., and Bedford, M. (1976). Space Syntax. Environment and Planning
Series B, 3: 14785.
Koch, D. (2015). ...and Avoidance. Proceedings 10th International Space Syntax Symposium, 22/1: 1
12.
Krenz, K. (2017). Regional Morphology: The Emergence of Spatial Scales in Urban Regions. Heitor
T., Serra M., Silva J. P., Bacharel M., and Silva L. C. da (eds) Proceedings of the 11th Space
Syntax Symposium, p. 74.174.23. Lisbon, Portugal: Instituto Superior Técnico, Departamento de
Engenharia Civil, Arquitetura e Georrecursos, Portugal.
Kuhn, T. S. (1962). The Structure of Scientific Revolutions. Chicago: University of Chicago Press.
Marshall, S., Gil, J., Kropf, K. et al. (2018). 'Street Network Studies: from Networks to Models and
their Representations'. Networks and Spatial Economics. https://doi.org/10.1007/s11067-018-
9427-9, 1–15
Netto, V. M. (2016). ‘What is space syntax not?’ Reflections on space syntax as sociospatial theory.
URBAN DESIGN International, 21/1: 2540. DOI: 10.1057/udi.2015.21
R Development Core Team. (2018). R: A language and environment for statistical computing. Vienna,
Austria: R Foundation for Statistical Computing.
Sabidussi, G. (1966). The centrality index of a graph. Psychometrika, 31/4: 581603. DOI:
10.1007/BF02289527
Saitta, L., and Zucker, J.-D. (2013). Abstraction in Artificial Intelligence and Complex Systems.
Abstraction in Artificial Intelligence and Complex Systems, Vol. 9781461470. New York, NY:
Springer New York. DOI: 10.1007/978-1-4614-7052-6
Turner, A. (2000). Angular Analysis: A Method For The Quantification Of Space ( No. 23). Working
Paper Series. London.
Ure, J. (1971). Lexical Density and Register Differentiation. Perren G. and Trim J. L. M. (eds)
Applications of Linguistics, pp. 44352. Cambridge University Press: London.
Wallentin, G. (2017). Spatial simulation: A spatial perspective on individual-based ecologya review.
Ecological Modelling, 350: 3041. Elsevier B.V. DOI: 10.1016/j.ecolmodel.2017.01.017
Westin, S. (2014). The Paradoxes of Planning: A Psycho-Analytical Perspective, 1st ed. London:
Routledge.
Zipf, G. K. (1932). Selected Studies of the Principle of Relative Frequency in Language. Cambridge,
Massachusetts: Harvard University Press.
18
Proceedings of the 12th Space Syntax Symposium
APPENDIX
List of all 287 identified terms and concepts:
A: accessibility, accessibility analysis, accessibility graph, affordance, agent, agent based,
aggregation, angular, angular analysis, angular betweenness, angular choice, angular depth, angular
distance, angular integration, angular segment, angular step, asymmetry, axial, axial analysis, axial
graph, axial integration, axial line, axial map, axial representation, axial structure.
B: background, background network, betweenness, betweenness.
C: centrality, centrality, centrality measure, centrality pattern, centrality value, choice, choice
analysis, choice map, choice measure, choice radius, choice value, closeness, closeness centrality, co
awareness, co presence, co present, co visibility, cognition, compactness, complex building, complex
system, configuration, configuration analysis, configurational, configurational measure,
configurational pattern, configurational property, connectivity, conservative, convex, convex analysis,
convex map, convex partitioning, convex space, convexity, copresence, core.
D: deformed grid, deformed wheel, depth, depth analysis, depth measure, depthmap, depthmapx,
description retrieval, distance decay, dual graph.
E: embeddedness, emergence, encounter, entropy, euclidean distance.
F: foreground, foreground network, foreground structure.
G: gate count, generative, generative process, generic, generic function, genotype, geometrical,
global, global accessibility, global choice, global integration, global measure, global movement,
global pattern, global radius, global scale, global structure, global system, globally integrated, graph
theory, grid condition, grid intensification, grid structure.
H: human behaviour.
I: inequality genotype, integrating, integration, integration analysis, integration core, integration map,
integration measure, integration value, intelligibility, intelligibility value, intelligible, intensification,
inter visibility, interface, interface map, isovist, isovist analysis, isovist area, isovist field, isovist
integration, isovist occlusivity.
J: justified, justified graph.
L: local, local accessibility, local centrality, local integration, local measure, local movement, local
structure, locally integrated.
M: macro scale, metric distance, metric integration, metric radius, micro scale, morphic language,
morphology, movement behaviour, movement economy, movement flow, movement network,
movement pattern, movement potential, multiplier effect.
N: nach value, nain value, natural movement, natural surveillance, network accessibility, network
analysis, network centrality, network configuration, network structure, node, node count, non
discursive, normalised angular, normalised integration.
O: occlusivity.
P: path, permeability, pervasive, place syntax, point isovist.
R: ra value, radius radius, relative asymmetry, residential activity, residential space, rra value.
S: scale network, segment analysis, segment angular, segment choice, segment connectivity, segment
integration, segment length, segment map, segment model, segregated, segregated space, segregation,
shallow space, shortest path, social encounter, social integration, social segregation, social solidarity,
social structure, solidarity, space occupancy, space syntax, spaciousness, spatial accessibility, spatial
behavior, spatial boundary, spatial centrality, spatial character, spatial cognition, spatial configuration,
spatial culture, spatial design, spatial hierarchy, spatial integration, spatial law, spatial logic, spatial
morphology, spatial order, spatial organisation, spatial organization, spatial practice, spatial process,
spatial proximity, spatial relation, spatial representation, spatial segregation, spatial structure, spatially
segregated, step depth, street configuration, street network, street pattern, street segment, structured,
symmetry, synergy, synergy value, syntactic, syntactic analysis, syntactic measure, syntactic model,
syntactic property, syntactic value, syntactical, syntactical analysis, syntactical property, syntax.
19
Proceedings of the 12th Space Syntax Symposium
T: topo geometric, topological, topological accessibility, topological choice, topological depth,
topological distance, topological radius, topological step, topological structure, total depth, transpatial,
tree structure.
U: urban configuration, urban grid, urban network.
V: vantage point, vga, vga analysis, virtual community, visibility, visibility analysis, visibility cost,
visibility graph, visibility polygon, visitor, visual access, visual accessibility, visual analysis, visual
connection, visual connectivity, visual control, visual depth, visual field, visual graph, visual
information, visual integration, visual relationship, visual step, visually integrated.
W: way finding, weak tie, weighted choice.
Tab l e 4: Scaled frequency of selected concepts per conference year.
Year
1997
1999
2001
2003
2005
2007
2009
2012
2013
2015
2017
metric_radius
0
0
0
0.067
0.348
1.583
1.486
1.399
0.612
0.527
1.598
angular_integration
0
0
0.176
0.448
0.440
0.375
0.591
0.853
1.327
2.033
1.569
integration
1.264
1.268
0.827
1.062
0.942
0.857
0.817
0.893
0.869
0.809
0.698
choice
0.301
0.267
0.810
0.342
0.612
0.948
1.037
1.154
1.342
1.397
1.277
segment
0.166
0.192
0.134
0.311
0.793
1.168
0.800
1.189
1.453
1.177
1.510
Tab l e 5: Scaled frequency of selected concepts per conference year.
Year
1997
1999
2001
2003
2005
2007
2009
2012
2013
2015
2017
betweenness centrality
0.247
0
0
0.062
0.053
0.136
0.350
0.320
1.439
1.476
2.332
axial map
1.551
0.977
1.155
0.925
1.045
0.978
0.817
0.669
0.832
0.571
0.520
topology
0.729
1.559
0.733
0.632
1.119
1.298
0.904
0.739
0.630
0.867
0.806
topological accessibility
0
0
0
0
0.202
0.172
0
0.348
0.457
1.262
2.830
closeness
0.054
0.133
0.036
0.108
0.548
1.040
0.759
1.197
0.695
1.811
1.677
occlusivity
0
0.176
0.193
0
0.442
0.270
0.476
0.222
0.149
1.141
2.840
visibility analysis
0
0.185
0
0
0.320
0.874
1.588
0.883
1.497
1.238
1.424
Tab l e 6: Scaled frequency of selected concepts per conference year.
Year
1997
1999
2001
2003
2005
2007
2009
2012
2013
2015
2017
affordance
0
0
0.066
1.172
1.296
0.295
0.869
0.647
1.021
1.432
1.608
agent based
0
0.381
0
2.458
0.331
0.846
0.382
0.342
0.847
0.827
1.150
entropy
0.100
0
0.402
0.992
1.668
0.713
0.552
1.141
0.502
1.482
1.226
spatial practice
0
0
0
0
1.498
0.365
0.442
2.118
0.887
1.276
0.724
20
Proceedings of the 12th Space Syntax Symposium
Tab l e 7: Scaled frequency of selected concepts per conference year.
Year
1997
1999
2001
2003
2005
2007
2009
2012
2013
2015
2017
spatial law
0
0
3.002
0.284
0.489
0
0.674
0.421
0.123
0.139
0.039
description retrieval
0.219
0
1.205
2.733
0.188
0
0.933
0
0.284
0.107
0.182
visibility
0.518
0.211
0.800
0.888
0.962
1.181
1.327
0.956
1.171
1.063
0.873
visual connection
0.915
1.133
0.420
0.685
1.705
0.838
0.325
0.903
0.543
1.117
1.077
virtual community
2.384
2.024
0.047
0.204
0.234
0.225
0.145
0.025
0.154
0.116
0.099
segregated space
1.864
0.256
1.353
1.602
0.623
0.493
0.184
0.804
0.402
0.531
0.559
movement economy
0.938
2.404
0.322
1.437
0.374
0.441
0.475
0.272
0.455
0.523
0.236
natural movement
1.960
1.486
0.660
1.018
0.499
0.630
0.554
0.548
0.531
0.488
0.839
human behaviour
0.149
0.369
0.410
0.968
1.282
0.601
0.953
0.331
1.256
1.420
1.456
spatial practice
0
0
0
0
1.498
0.365
0.442
2.118
0.887
1.276
0.724
!
21
ResearchGate has not been able to resolve any citations for this publication.
Conference Paper
Full-text available
Joel Lamere's prompt for this session of the 2019 ACSA Conference asks whether there are metrics other than the calculus of curved surfaces for making architecture. There are, and isovist theory provides a family of them. Developed continuously since 1979, isovist theory (and its 'cousin,' space syntax) has accumulated a considerable literature. With advances in computing power and coding techniques, isovist theory is poised to enter both design and design research processes fluidly and in a visually arresting way. This paper reviews the basics of isovist theory, discusses the software application ISOVIST, and concludes by discussing how isovists and isovist fields represent a new approach to deploying geometry's experiential/relational power, to "metricising" architectural space, and to understanding what space is.
Book
Full-text available
Using a philosophical and psychoanalytical approach, this book critically examines expert knowledge within architecture and urban planning. Its point of departure is the 'gap' between visions and realities, intentions and outcomes in planning, with particular focus on projects in Sweden that try to create an 'urban atmosphere'. Finding insights from the work of Sigmund Freud and his followers, the book argues that urban planning during the 20th century is a neurotic activity prone to produce a type of alienation.
Article
Full-text available
Over the last fifty years, research into street networks has gained prominence with a rapidly growing number of studies across disparate disciplines. These studies investigate a wide range of phenomena using a wealth of data and diverse analytical techniques. Starting within the fields of transport or infrastructure engineering, street networks have commonly been treated as sets of more or less homogeneous linear elements, connecting locations and intersecting at junctions. This view is commonly represented as a graph, which provides a common and rigorous formalisation accessible across disciplines and is particularly well-suited for problems such as flow optimisation and routing. Street networks are, however, complex objects of investigation and the way we model and then represent them as graphs has fundamental effects on the outcomes of a study. Many approaches to modelling street networks have been proposed, each lending itself to different analyses and supporting insights into diverse aspects of the urban system. Yet, this plurality and the relation between different models remains relatively obscure and unexplored. The motivations for adopting a given model of the network are also not always clear and often seem to follow disciplinary traditions. This paper provides an overview of key street network models and the prima facie merits of pertinent alternative approaches. It suggests greater attention to consistent use of terms and concepts, of graph representations and practical applications, and concludes with suggestions for possible ways forward.
Article
Full-text available
quanteda is an R package providing a comprehensive workflow and toolkit for natural language processing tasks such as corpus management, tokenization, analysis, and visualization. It has extensive functions for applying dictionary analysis, exploring texts using keywords-in-context, computing document and feature similarities, and discovering multi-word expressions through collocation scoring. Based entirely on sparse operations, it provides highly efficient methods for compiling document-feature matrices and for manipulating these or using them in further quantitative analysis. Using C++ and multithreading extensively, quanteda is also considerably faster and more efficient than other R and Python ackages in processing large textual data. The package is designed for R users needing to apply natural language processing to texts, from documents to final analysis. Its capabilities match or exceed those provided in many end-user software applications, many of which are expensive and not open source. The package is therefore of great benefit to researchers, students, and other analysts with fewer financial resources. While using quanteda requires R programming knowledge, its API is designed to enable powerful, efficient analysis with a minimum of steps. By emphasizing consistent design, furthermore, quanteda lowers the barriers to learning and using NLP and quantitative text analysis even for proficient R programmers.
Article
Spatial Simulation is a spatially explicit, bottom-up modelling approach that includes individual-based models and cellular automata. While spatial heterogeneity and individual variation have been considered as noise in the past, this is exactly what has become the centre of interest of the individual-based paradigm in ecology. According to Individual-based Ecology, the interaction and behaviour of individual organisms leads to the emergence of macro-level patterns on the system scale. Although individual-based models have almost always been spatially explicit, the focus was commonly given to temporal processes and behavioural rules over spatial aspects. Today, the wide availability of spatial data and ever increasing computational power together with a strive for realistic models has renewed the attention to spatial aspects in simulation modelling. This review provides an overview of the state of the art of Spatial Simulation modelling in Ecology, reviews its limitations and open issues and it discusses future research avenues by taking an explicit geospatial perspective. The main avenues that are discussed revolve around the role of spatial context to determine the structure of living systems, potentials of hybrid top-down/bottom-up model designs to integrate hierarchical, spatial and temporal scales of ecological systems, and current trends in the representation and analysis of simulated spatio-temporal (big) data.
Scitation is the online home of leading journals and conference proceedings from AIP Publishing and AIP Member Societies