THE INTERNATIONAL ENCYCLOPEDIA OF GEOGRAPHY: PEOPLE,

THE EARTH, ENVIRONMENT, AND TECHNOLOGY

Human-centered Geospatial Data Science

Yuhao Kanga
a
GISense Lab, Department of Geography and the Environment, The University of Texas at
Austin, TX, United States; yuhao.kang@austin.utexas.edu

ARTICLE HISTORY
arXiv:2501.05595v1 [cs.CY] 9 Jan 2025

Compiled January 13, 2025

ABSTRACT
This entry provides an overview of Human-centered Geospatial Data Science, high-
lighting the gaps it aims to bridge, its significance, and its key topics and research.
Geospatial Data Science, which derives geographic knowledge and insights from
large volumes of geospatial big data using advanced Geospatial Artificial Intelli-
gence (GeoAI), has been widely used to tackle a wide range of geographic problems.
However, it often overlooks the subjective human experiences that fundamentally
influence human-environment interactions, and few strategies have been developed
to ensure that these technologies follow ethical guidelines and prioritize human val-
ues. Human-centered Geospatial Data Science advocates for two primary focuses.
First, it advances our understanding of human-environment interactions by lever-
aging Geospatial Data Science to measure and analyze human subjective experi-
ences at place including emotion, perception, cognition, and creativity. Second, it
advocates for the development of responsible and ethical Geospatial Data Science
methods that protect geoprivacy, enhance fairness and reduce bias, and improve the
explainability and transparency of geospatial technologies. With these two missions,
Human-centered Geospatial Data Sciences brings a fresh perspective to develop and
utilize geospatial technologies that positively impact society and benefit human well-
being and the humanities.

KEYWORDS
Human-centered Geospatial Data Science; place; ethics; geospatial big data; GeoAI

1. Introduction

The complex relationships and interactions between humans and their natural en-
vironment have long been a central topic in geography, a focus that dates back to
the Man-Land Tradition (Pattison 1964). Geographers have studied the dual rela-
tionship between humans and the natural world, examining how humans adapt to,
modify, and affect their surroundings, and conversely, how the environment shapes
human activities and cultural practices (Tuan 1974). The introduction of Geographic
Information Systems (GIS) and, more recently, Geospatial Data Science has pro-
vided unprecedented opportunities to understand human-environment relationships
through advanced data collection methods and enhanced spatial analysis and model-
ing. Geospatial Data Science, the discipline of extracting geographic knowledge from

CONTACT Yuhao Kang. Email: yuhao.kang@austin.utexas.edu

geospatial big data, utilizing advanced geocomputational methods such as Geospa-
tial Artificial Intelligence (GeoAI) (Janowicz et al. 2020), could significantly enhance
our understanding of human-environmental interactions. Specifically, the emergence of
geospatial big data including large-scale human mobility data, user-generated social
media content, extensive street view images, and remote sensing images, offers solid
data foundations for digitally representing and understanding the world from a data-
driven perspective. Moreover, advanced geocomputational techniques, ranging from
GIS to cutting-edge Geospatial Artificial Intelligence (GeoAI), allows for the effec-
tive handleing of complex geographical tasks, such as spatial reasoning, geographical
phenomena modeling, and geospatial analysis and simulations. Numerous applications
utilizing GIS and Geospatial Data Science have been developed, benefiting various do-
mains ranging from large-scale land use monitoring, effective resource management,
emergency response and hazard mapping, to smart city planning, etc. Despite their
success, researchers have observed two challenges within the current GIS and Geospa-
tial Data Science.
First, the current development and applications of GIS and Geospatial Data Science
have primarily been driven by technology, and have paid relatively insufficient atten-
tion to human experience. Geographic phenomena are often treated as functions of
physical settings such as terrain, climate, and infrastructure, while neglecting the sub-
jective experiences of individuals and communities. Contemporary research in GIS and
Geospatial Data Science often adhere to static spatial notions (Kwan 2013), concen-
trating on the tangible properties of space (Freundschuh and Egenhofer 1997; Giordano
and Cole 2018). This perspective overlooks the dynamic and complex ways in which
humans interact with and perceive their environments (Shaw and Sui 2021), poten-
tially missing the nuanced and varied subjective experiences of different individuals
and groups within the physical space.
Humans experience place rather than abstract space. Beyond space, place is infused
with human meanings, encompassing individual social connections, attitudes, and val-
ues, all rooted in the subjective nature of human experience. People learn about and
interpret their surroundings by utilizing their senses, such as vision, touch, and hearing,
forming what is known as sense of place (Tuan 1977). These senses guide our behav-
iors and shape our interactions with our environments, reflecting our mental space
(Shaw and Sui 2021). Consider the experience of coming home after an exhausting
day at work. Your home, located in a neighborhood perceived as safe and welcom-
ing, including familiar streets and friendly neighbors, may evoke positive emotions of
comfort and a sense of belonging, fostering meaningful and supportive environments.
What constitutes human subjective experiences such as emotions, perceptions, cog-
nition, and creativity toward various environmental settings? Leveraging Geospatial
Data Science not only helps answer this question but also enables researchers to gain a
deeper understanding of the diverse ways people perceive and experience their environ-
ments, leading to more inclusive and effective spatial decision making. Furthermore,
enriching GIS and Geospatial Data Science with human dimensions can benefit the
development of these technologies, ensuring they are more closely aligned with human
behaviors (Zhao 2022).
Second, the social and ethical implications of GIS and Geospatial Data Science
has received insufficient attention (Crampton 1995). The use and potential misuse of
Geospatial Data Science may pose several challenging ethical issues and raise deep
concerns related to geoprivacy, bias and fairness, and explainability and transparency
(Pavlovskaya 2018; Nelson et al. 2022). For instance, large-scale human mobility data
can reveal sensitive personal information, such as home addresses and frequently vis-

2
 Figure 1. A conceptual framework of Human-centered Geospatial Data Science

ited locations, emphasizing the crucial need to protect this information from misuse
and unauthorized access. Using biased data for training crime prediction models may
lead to increased surveillance and discrimination against certain populations, perpet-
uating cycles of inequality (Mayson 2018). Additionally, algorithms used to support
decision-making, such as resource allocation, need to be transparent so that stakehold-
ers can understand how decisions are made and trust the outcomes. However, limited
strategies have been developed to address these social and ethical issues associated
with Geospatial Data Science. Questions arise such as where is the human element in
these technologies, and who are the humans behind these technologies? Developing eth-
ical GIScience technology involves prioritizing fundamental human values—fairness,
transparency, privacy, safety, trust, inclusivity, sustainability, and legal compliance—
to ensure technological innovation respects human rights. Incorporating these human-
centered social and ethical considerations into the development of Geospatial Data
Science supports more equitable and inclusive design and decision-making, ultimately
creating environments that enhance human life and social well-being.
Human-centered Geospatial Data Science aims to understand human experiences
and prioritize human values within the field of Geospatial Data Science. By adopting a
more holistic and people-oriented understanding of geography, this framework bridges
critical gaps by focusing on two main missions: (1) advancing human-environment re-
lationships through a deep understanding of human experiences, and (2) developing
trustworthy and responsible geospatial technologies that prioritize human values. Fig-
ure 1 shows an overview of the conceptualization of Human-centered Geospatial Data
Science.

2. Understanding Human Experiences

Human experiences of place consist of a broad spectrum of internal mental processes,

such as emotion, perception, cognition, and creativity. Traditionally, to collect these

3
human dimensions of experiences, researchers have to rely on questionnaires, surveys,
interviews, etc., which are labor-intensive and time-consuming. Additionally, the in-
herently subjective nature of these experiences poses significant challenges for accurate
measurement and analysis. Geospatial Data Science, leveraging multi-source geospa-
tial big data and advanced GeoAI methods, offers new opportunities for understanding
human subjective experiences.

2.1. Emotion
Different places could evoke a variety of human emotions, deeply influencing our inti-
mate bodily states. When people interact with various places or events, their emotional
responses might be affected by the environmental settings of places (Feng et al. 2024).
For example, a child is likely to feel joy and excitement at Disneyland; while we may feel
a sense of solemnity or sadness when visiting a memorial site. By utilizing Geospa-
tial Data Science, researchers can now map emotional landscapes and analyze how
geographical factors influence our emotions. Geotagged user-generated content has
offered new avenues for examining human emotions at places. People often express
their emotions, both consciously and unconsciously, at various social media platforms.
Using advanced GeoAI methods, researchers can tap into this rich data source to mea-
sure human emotions associated with specific places. Prior studies can be primarily
categorized into two streams:
First, text-based analytical methods utilizes massive geotagged social media tex-
tual content and applies Natural Language Processing (NLP) techniques to decipher
and analyze the emotions expressed by people. Researchers can measure multiple di-
mensions of human emotions, including happiness, sadness, and positive or negative
sentiments, by examining tweets, comments, and posts, This provides valuable insights
into fluctuations of public sentiment and mood across different geographical spaces.
Another method refers to image-based analytical methods. Using computer vision
techniques, researchers could extract and analyze human emotions from facial expres-
sions captured in images. From selfies shared on social media to crowd photos at
events, this technique allows for the assessment of emotional responses in situ, offering
a direct measure of how people emotionally react to the environments.
In addition to these two primary research streams, some emerging psychophysi-
ological technologies, such as functional magnetic resonance imaging (fMRI), elec-
troencephalogram (EEG), electrocardiogram (ECG), electrodermal activity (EDA),
and electromyography (EMG) offer new measuring methods for understanding human
emotions (Zhao et al. 2024). By associating with real-world or virtual environments,
researchers could leverage these methods to better model human emotion-environment
interactions.

2.2. Perception
Human perception of place describes how individuals experience their environment
through various sensory dimensions such as sight, sound, and smell. The built envi-
ronment plays a crucial role in shaping human perceptions by affecting visual, audi-
tory, and other experiences (Ito et al. 2024). For instance, a neighborhood with quiet
streets, clear sounds of birds, and visually appealing architecture is likely to be per-
ceived as safe and aesthetically pleasing by its residents. This reflects the complex
human-environment interactions, highlighting how environmental factors influence in-

4
dividual perceptions and feelings. To measure human perceptions of place, two types
of geospatial big data have been widely utilized.
First, geotagged user-generated content posted on social media platforms often re-
flects people’s perceptions of specific places or events. These digital footprints could
reveal how individuals experience and interact with various surrounding environments,
reflected through the content they share, such as photos, videos, and texts. Analyzing
geotagged social media data can reveal trends and patterns of multiple dimensions
of human perceptions, such as human visual, auditory, and olfactory perceptions at
different places.
In recent years, street view imagery has provided another valuable data source for
assessing human perceptions of the built environment. Street view images capture
detailed streetscape settings at eye level, providing a realistic perspective of how in-
dividuals might perceive their surrounding environment. Then, researchers leveraged
advanced computer vision approaches, such as Convolutional Neural Networks (CNNs)
to measure human perceptions of place in response to street view images. The hypoth-
esis is that people’s subjective perceptions of the built environment captured in street
view images can be assessed to reflect their place perceptions. Researchers have utilized
such a computational workflow and measured multiple dimensions of human percep-
tions of place including safe, beautiful, depressing, and lively. This method enables
researchers to understand the nuanced subjective human experiences, and advance
our understanding of human-environment interactions.

2.3. Cognition
Human cognition of place reflects the psychological processes of how people conceptu-
alize, interpret, and make sense of geographical environments. Through synthesizing
multi-sensory perceptions of the environment based on past experiences and prior
knowledge, individuals interact with their surrounding environments, fostering their
place identity, awareness of vague place, and sense of direction and distance. These
cognitive processes could influence key behaviors, such as navigation, planning, and
daily decision-making. Notably, the way in which places are conceptualized can vary
significantly among individuals, influenced by the differences in their cultural back-
ground, personal experiences, and emotional connections. These factors may lead to
potentially diverse cognitive processes and interpretations, even for the same place,
highlighting the challenges in incorporating subjective cognition of place into Geospa-
tial Data Science. To collect and analyze human cognition of place, multi-sourced social
media data has become a valuable data source that reflects how people conceptual-
ize and perceive different environments. Researchers have leveraged social media data
to capture real-time expressions and analyze human mental space, thereby enriching
our understanding of place identity, vague cognitive regions, toponyms, and perceived
distance and orientation. For instance, by leveraging crowdsourced social media data,
researchers could explore the unique characteristics that contribute to place identity
based on associated texts describing places in social media posts. Researchers have
identified and characterized vague cognitive regions and place names—areas without
clear physical boundaries but are frequently referenced in the daily lives of individu-
als. Researchers have also examined the sense of distance and sense of orientation to
understand how people navigate and make spatial decisions. Having a deeper under-
standing of human cognition of place using social media data could benefit various
applications in spatial decision-making. We could understand the complete ways how

5
people conceptualize place, leading to potential effective urban planning and personal
navigation solutions.

2.4. Creativity
Human creativity — the ability to generate original ideas and expressions — is another
important component of human experience. Artists and creators take the inputs from
human experiences and adapt them into compelling narratives and innovative solutions
(Jung 2020). In Geospatial Data Science and GIS, the outcomes of human creativity
are often embodied in the art of map design and geovisualizations, which are used for
creating visually pleasing maps and conveying geographic information through visual
storytelling (Kang et al. 2024). Traditionally, it has been believed that machines are
unable to replicate the artistic and emotional elements of cartographers’ creativity
and experience due to the inherently subjective nature of artistic expression. However,
advancements in GeoAI methods may revolutionize this field by enabling the model-
ing of various aspects of map design, such as map styles, visual variables, and map
critique. Moreover, the recent emergence of Generative Artificial Intelligence (GenAI)
has demonstrated exceptional promise in stimulating and enhancing map creativity
and aesthetics. Researchers have been leveraging text-to-image models such as Stable
Diffusion, DALL-E 2, and Midjourney not only to generate maps directly but also to
foster innovative and creative ideas that inspire map design. Additionally, text-to-text
models like ChatGPT are utilized to automate the map-making process. These devel-
opments in AI technologies are opening up new possibilities for understanding human
subjectivity and creativity, potentially changing the future paradigm of map-making
and expanding what we can visualize.

3. Prioritize Human Values with Ethical Discussions and Practices

The ethics of GIS and Geospatial Data Science refers to a set of guidelines and prin-
ciples designed to ensure the alignment of geospatial technologies and methodologies
with human values and uphold human rights. This ethical focus is crucial for guid-
ing the responsible use and development of trustworthy geospatial technologies. Re-
searchers from diverse academic and professional communities including GIScience,
geography, and computer science have provided their perspectives on the ethical dis-
cussions in geospatial theories and practices. However, few strategies have been de-
veloped to implement these ethical considerations in geospatial practices (Kang et al.
2024). Despite the complexity of ethical issues, the following subsections outline three
main aspects of ethics in geospatial data science: geoprivacy, bias and fairness, and
explainability and transparency. (Ilägcrstrand 1970) Several recent advancements in
tackling ethical issues and developing responsible Geospatial Data Science practices
are summarized.

3.1. Geoprivacy
Human geoprivacy refers to the right and ability of individuals to govern the collection,
sharing, and use of their location-based information. It protects people’s locational
privacy from unauthorized access and prevents potential misuse of their sensitive geo-
graphical data. In our daily lives, we frequently interact with devices that are equipped

6
with GPS and make use of location-based services. Activities such as sharing geotagged
posts on social media platforms, navigating with GPS-enabled devices, and granting
mobile apps permission to access our location data can unintentionally disclose our
personal information to different entities, including governments and private sectors.
The data that is shared can expose details about our trajectories, activities, and be-
haviors, emphasizing the crucial need for effective strategies to protect geoprivacy.
To address these concerns, researchers have investigated multiple methods to en-
hance human geoprivacy in the existing multiple large-scale geospatial datasets. For ex-
ample, data aggregation and geomasking techniques have been widely used to obscure
location information to enhance human geoprivacy. Additionally, advanced GeoAI
methods have been employed to further enhance the protection of geoprivacy. Genera-
tive Adversarial Networks (GANs), for instance, have been utilized to simulate human
trajectories and generate synthetic datasets, allowing researchers to analyze patterns
without accessing sensitive individual raw data. Another cutting-edge approach, fed-
erated learning, enables the training of AI models on multiple decentralized devices
that store local datasets. This method ensures that the real data remains on the de-
vice and only changes to the model are shared across the network, thus protecting
individual privacy. By employing these advanced techniques for protecting geoprivacy,
researchers can ensure that geospatial data management is not only responsible and
ethical but also cultivating trust to promote the beneficial uses of Geospatial Data
Science.

3.2. Bias and Fairness

A fundamental principle in Human-centered Geospatial Data Science is recognizing
the humans behind technologies and understand how they engage with these systems.
Addressing bias and fairness in Geospatial Data Science is crucial to prevent potential
discrimination against any group or individual and to guarantee that outcomes from
geospatial technologies are both fair and equitable. There are two forms of bias com-
monly seen in current Geospatial Data Science studies. Population bias relates to how
demographic characteristics such as age, gender, race and ethnicity, are represented
in geospatial data. Given the individual differences in subjective human experiences
and the collection methods employed, datasets frequently become skewed, and favor-
ing certain demographic groups over others. For example, geospatial data collected
from mobile apps typically overrepresent younger urban users, while underrepresent-
ing older rural populations. This discrepancy can lead to models that fail to accurately
capture the behaviors or needs of the entire population, potentially leading to biased
outcomes and discriminatory practices within Geospatial Data Science. Spatial bias,
on the other hand, concerns the geographic representativeness within data. Geospa-
tial data often is not collected uniformly across all regions, leading to discrepancies
in coverage that distort the analysis and applications of the data. For instance, ur-
ban areas and developed countries generally have more extensive data coverage. While
their counterparts may suffer from significant underrepresentation. This uneven spa-
tial coverage can skew analysis and modeling, resulting in geographic disparities in the
accuracy of data and efficiency of geospatial applications.
To enhance fairness and reduce bias in Geospatial Data Science, researchers could
focus on both the data collection phase and during the development of algorithms to
tackle these challenges. In the data collection phase, researchers have fused multiple
data sources to diversify data sources. Each data source has its own pros and cons.

7
Leveraging multi-source geospatial big data may cover a broader spectrum of geo-
graphic and demographic characteristics, ensuring a more balanced representation of
different areas and populations. Moreover, increasing attention has been paid to cer-
tain underrepresented and vulnerable population groups and areas, which historically
may have been overlooked. Through specifically targeting these groups, researchers
could enhance the diversity and equity of the collected datasets. Additionally, the suc-
cess of Public Participatory GIS (PPGIS) has highlighted the benefits of community
engagement in geospatial projects. Given the nature of spatial heterogeneity, Geospa-
tial Data Science need to account for local contexts. Involving local communities in
the data collection process can enrich Geospatial Data Science with their unique in-
sights and local knowledge. This collaborative process not only enhances the richness
and accuracy of the collected datasets but also fosters greater trust and inclusivity
in the data collection process. During the algorithm development phase, researchers
have explored several bias-mitigation algorithms designed to address and neutralize
biases within GeoAI models. By adjusting weights during the training process, these
algorithms may reduce biases and enhance fairness. These strategies provided possible
solutions for addressing bias and promoting fairness in Geospatial Data Science to
ensure inclusive practices.

3.3. Transparency and Explainability

The current GeoAI methods utilized in Geospatial Data Science, ranging from ma-
chine learning, deep learning, and the emerging Generative AI (GenAI), often face
criticism for their “black-box” nature. Thus, the lack of transparency in these models
makes it difficult to understand their underlying mechanisms, which in turn hinders
human trust. Enhancing transparency and explainability in Geospatial Data Science
involves the ability to decompose, understand, track, and effectively communicate the
decision-making processes of GeoAI methods in modeling geographic phenomena. It is
crucial to articulate how these models derive their predictions in ways that users can
trust and validate. This ensures that stakeholders can make informed decisions in a
responsible way. To advance transparency and explainability in Geospatial Data Sci-
ence, researchers have explored several innovative approaches. For instance, researchers
have placed more emphasis on the development of eXplainable AI (XAI) methods to
understand and interpret the complex internal processes of AI models. These meth-
ods could show which variables and how they influence model decisions. Additionally,
researchers have also leveraged advanced visualization strategies to effectively explain
the model outputs. Both strategies could not just deepen our understanding of GeoAI
models, but also provide a robust foundation of trust and responsibility of geospatial
technologies. These efforts ensure that GeoAI models are comprehensible and trans-
parent, fostering their alignment with societal needs and expectations and promoting
their responsible use.

4. Conclusions

Human-centered Geospatial Data Science is designed to augment human intelligence

and benefit society. Using large-scale geospatial big data and advanced GeoAI, it fo-
cuses on understanding how humans interact with the surrounding environments and
prioritizing human values in both research and practices of Geospatial Data Science.
By enriching Geospatial Data Science with human subjective experiences, researchers

8
could advance the knowledge of multiple human dimensions including emotion, per-
ception, and cognition of place, and creativity in mapmaking. Such an approach helps
understand the complex human-environment interactions and could inform the de-
velopment human-centered algorithms. To build responsible intelligent systems, it is
crucial to understand and prioritize human values such as protecting human geopri-
vacy, reducing bias and enhancing fairness, as well as improving transparency and
explainability of geospatial technologies. These efforts are essential for ethical use and
gaining user trust. Looking beyond, the development of geospatial technologies needs
to be guided by human-centric principles to prioritize the needs and well-being of hu-
mans. This will ensure they are not only innovative but also have positive impacts
across all sectors of society.

Cross-references

See Also: GeoAI and Its Implications, Social Sensing, Data Science, GeoAI and Deep
Learning, Platial, Spatial Data Science

References

Crampton, J. (1995). The ethics of GIS. Cartography and geographic information systems,
22(1):84–89.
Feng, J., Zhao, B., and Shaw, S.-L. (2024). The study of emotion in GIScience: status and
prospects. Cartography and Geographic Information Science, pages 1–16.
Freundschuh, S. M. and Egenhofer, M. J. (1997). Human conceptions of spaces: implications
for GIS. Transactions in GIS, 2(4):361–375.
Giordano, A. and Cole, T. (2018). The limits of GIS: Towards a GIS of place. Transactions
in GIS, 22(3):664–676.
Ilägcrstrand, T. (1970). What about people in regional science. regional science association,
24.
Ito, K., Kang, Y., Zhang, Y., Zhang, F., and Biljecki, F. (2024). Understanding urban percep-
tion with visual data: A systematic review. Cities, 152:105169.
Janowicz, K., Gao, S., McKenzie, G., Hu, Y., and Bhaduri, B. (2020). GeoAI: spatially explicit
artificial intelligence techniques for geographic knowledge discovery and beyond.
Jung, J.-K. (2020). Teaching creative geovisualization: Imagining the creative in/of GIS. The
Canadian Geographer/Le Géographe Canadien, 64(4):512–528.
Kang, Y., Gao, S., and Roth, R. E. (2024). Artificial intelligence studies in cartography: a
review and synthesis of methods, applications, and ethics. Cartography and Geographic
Information Science, pages 1–32.
Kwan, M.-P. (2013). Beyond space (as we knew it): Toward temporally integrated geographies
of segregation, health, and accessibility: Space–time integration in geography and GIScience.
Annals of the Association of American Geographers, 103(5):1078–1086.
Mayson, S. G. (2018). Bias in, bias out. Yale Law Journal, 128:2218.
Nelson, T., Goodchild, M., and Wright, D. (2022). Accelerating ethics, empathy, and eq-
uity in geographic information science. Proceedings of the National Academy of Sciences,
119(19):e2119967119.
Pattison, W. D. (1964). The four traditions of geography. Journal of Geography, 63(5):211–216.
Pavlovskaya, M. (2018). Critical gis as a tool for social transformation. The Canadian Geog-
rapher/Le Géographe Canadien, 62(1):40–54.
Shaw, S.-L. and Sui, D. (2021). Understanding the new human dynamics in smart spaces and
places: Toward a splatial framework. In Smart Spaces and Places, pages 7–16. Routledge.

9
Tuan, Y.-F. (1974). Topophilia: A study of environmental perception, attitudes, and values.
Columbia University Press.
Tuan, Y.-F. (1977). Space and place: The perspective of experience. U of Minnesota P.
Zhao, B. (2022). Humanistic GIS: Toward a research agenda. Annals of the American Asso-
ciation of Geographers, 112(6):1576–1592.
Zhao, B., Feng, J., Sun, Y., Chang, X., and Shaw, S.-L. (2024). Neural sensing: Toward a new
approach to understanding emotional responses to place. Transactions in GIS.

Further Readings

Thatcher, J., Shears, A., and Eckert, J. (2018). Thinking Big data in geography: New
regimes, new research. U of Nebraska Press.362
Williams, S., (2022). Data action: Using data for public good. MIT Press.
Shneiderman, B., 2022. Human-centered AI. Oxford University Press.

10