1 of 31

Welcome to GEOGM0068:

Geographic Information Retrieval and Integration

Rui Zhu

rui.zhu@bristol.ac.uk

GEOGM0068 - TB2 2024/25

2 of 31

Assessment

Deadline extended to April 3rd
Don’t wait till the last few days to complete it !!!!!
I am only available to answer questions by March 27th (Thursday). Do not expect me to reply your email or set up extra meetings after that date regarding your assessment !!!
DBpedia Spotlight’s server has issues these days. If you have troubles implementing it as a geoparser, try Geoparser library: https://docs.geoparser.app/en/latest/usage.html (or any other geoparsers you might find. There are many on the web!)

GEOGM0068 - TB2 2024/25

3 of 31

Lecture 10

Rui Zhu

rui.zhu@bristol.ac.uk

GEOGM0068 - TB2 2024/25

4 of 31

(Geo)Ethics

GEOGM0068 - TB2 2024/25

5 of 31

(Geo)Ethics - A Little Bit about Its History

Defined by Aristotle (384/383 B.C. – 322 B.C.) as the investigation and reflection on the operational behavior of humans, searching for legitimate criteria by which to evaluate behaviour and choices, and identifies that part of philosophy dealing with the problem to take decisions by the human agent.
In the early 90s, the word “Geoethics” began to be used to define the ethical and social implications of geosciences. The need to increase awareness of the ethical obligations of geoscientists' activity was formalised in 2014.
(Geo)Ethics in the context of (spatial) data science emerges around 2020.

GEOGM0068 - TB2 2024/25

6 of 31

Kranzberg’s Law

“technology is neither good nor bad; nor is it neutral”

(Kranzberg, 1986)

https://thefrailestthing.com/2011/08/25/kranzbergs-six-laws-of-technology-a-metaphor-and-a-story/

GEOGM0068 - TB2 2024/25

7 of 31

Power-Knowledge

“power produces knowledge…there is no power relation without the correlative constitution of a field of knowledge, nor any knowledge that does not presuppose and constitute at the same time power relations”

(Foucault, 1977)

GEOGM0068 - TB2 2024/25

8 of 31

Global Coverage of Geotagged Tweets

Leetaru, K., Wang, S., Cao, G., Padmanabhan, A., & Shook, E. (2013). Mapping the global Twitter heartbeat: The geography of Twitter. First Monday .

All Exact Location coordinates in the Twitter Decahose 23 October 2012 to 30 November 2012

GEOGM0068 - TB2 2024/25

9 of 31

Data Politics

Critical ethics helps us to better understand data politics
“A critical ethics is to “strike at oneself” in order to question the taken-for-granted”
A form of “doing critique” - not something that can only be learned by reading or knowing, but must be experienced

http://gistbok.ucgis.org/bok-topics/gis-and-critical-ethics

GEOGM0068 - TB2 2024/25

10 of 31

Two Concepts on (Geo)Ethics

Troublesome Knowledge

The idea that as we learn we encounter knowledge that is “difficult” for us in some way

Something that we’d much rather ignore or pretend isn’t true
It comes (or appears to come) from somewhere different from us, and it disrupts us.

Threshold Concept

It will transform the thinker, or take them to new places (hence, to lead them astray). It too, may be uncomfortable and involve a certain degree of unlearning

Different from core concepts such as scale, projections, spatial relations, etc., it is transformative, irreversible, integrative, and bounded

GEOGM0068 - TB2 2024/25

11 of 31

Troublesome Knowledge in (Spatial) Data

Census categories: Why such a category? Who made it? …
Your database is often full of numbers, lacking of meanings

GEOGM0068 - TB2 2024/25

12 of 31

Threshold Concept in (Spatial) Data

Power of Map: Is the representation of a map creative or passive? (Harley 1988)
“The real world is not completely symbolizable” is a harder concept

(such a threshold concept of the power of the map transforms our thinking of mapping from passively reflecting meanings to actively creating meaning)

GEOGM0068 - TB2 2024/25

13 of 31

Geoprivacy - A Key topic in GeoEthics

Any (geo)privacy concerns you find from this scenario?

GEOGM0068 - TB2 2024/25

14 of 31

Geoprivacy - A Key topic in GeoEthics

Any (geo)privacy concerns you find from this scenario?

GEOGM0068 - TB2 2024/25

15 of 31

Geoprivacy - A Key topic in GeoEthics

GEOGM0068 - TB2 2024/25

16 of 31

Again, Spatial is Special

Ubiquitous positioning devices and easy-to-use APIs make information about an individual’s location much easier to capture than other kinds of personally identifiable information
Users of information services have a substantial incentive to share their location with service providers, as location information can significantly improve the quality of a service and make it more useful.

GEOGM0068 - TB2 2024/25

17 of 31

Again, Spatial is Special

Users often share their current location unknowingly
Location-based inferences can reveal information that the user never intended or agreed to share with a service.
Knowing a customer’s location is an economic asset for a business

GEOGM0068 - TB2 2024/25

18 of 31

Again, Spatial is Special

Preserving geoprivacy involves more than obfuscating geographic coordinates. Location can be inferred from non-explicit geospatial information such as interests, activities, and socio-demographics.

I am not sharing any locational information on my Twitter account or Facebook, but I liked tweets/pages about the UoB, used tagges such as #avonriver #harborfestival #spatialdatascience …
How difficult is it to predict my location?

GEOGM0068 - TB2 2024/25

19 of 31

What Shall I Do in GIR?

AWARENESS!
Other strategies/techniques:

Location masking
Reproducibility and replicability
FAIR principle (Finability, Accessibility, Interpretability, and Reusability)
Data debiasing
Responsible AI
…

GEOGM0068 - TB2 2024/25

20 of 31

Responsible (Geo)AI

Be critical on what is fundamentally “artificial intelligence”, particularly how different it is to human intelligence

How much does language associated with other aspects of cognition (e.g. common sense; reasoning, etc)?
individual intelligence is deeply reliant on one’s participation in social and cultural environment
Avoid “anthropomorphizing” AI

Be aware of the harms (Geo)AI could bring to our society and environment

Carbon emission of training the model
Hallucination
Change to our learning habitat
Fairness, justice, and trust of our society

GEOGM0068 - TB2 2024/25

21 of 31

Summary

GEOGM0068 - TB2 2024/25

22 of 31

What Have We Learnt?

Concepts

Spatial data models
Types of spatial data
Spatial data interoperability
Georeferencing: geoparsing and geocoding
Gazetteers
Spatial indexing (Space-filling curve, Quad tree, RTree)
Spatial ranking (spatial relevance/similarity)
Geospatial semantics (geospatial ontology)
Semantic Web (geospatial knowledge graphs)
Expert-driven and data-driven approaches
GeoEthics

Spatial is Special!

GEOGM0068 - TB2 2024/25

23 of 31

What Have We Learnt?

Techniques

Relational (spatial) database management (e.g., spatial join, query, set CRS, etc.)
API (data collection)
Natural language processing (and its adoption in geography)
Place name identification and disambiguation
Bag-of-word approach
Evaluation metrics (precision, recall, and F-score)
TF-IDF
Geospatial knowledge graphs and ontologies
Word2Vec (CBOW and skip-gram)
WordCloud
Topic modeling

GEOGM0068 - TB2 2024/25

24 of 31

What Have We Learnt?

Tools

pandas / geopandas
spaCy
DBpedia Spotlight (spacy-dbpedia-spotlight), OpenStreetMap (osmnx / pyrosm), Foursquare API
wordcloud
sklearn
gensim
pyLDAvis
Matplotlib
rdflib
…

GEOGM0068 - TB2 2024/25

25 of 31

Geographic Data Science

GEOGM0068 - TB2 2024/25

26 of 31

Career

GEOGM0068 - TB2 2024/25

27 of 31

Academia vs Industry

GEOGM0068 - TB2 2024/25

28 of 31

Jobs You Can Look For

(Spatial) Data Scientist
(Spatial) Data Engineer
(Spatial) Data Curator
(Spatial) Data Analyst / Quantitative Analyst
Cartographer
Product Manager
Journalist (they do need data scientists)
Consultant
Banker
Government agent
Research assistant/associate
…

GEOGM0068 - TB2 2024/25

29 of 31

GEOGM0068 - TB2 2024/25

30 of 31

GEOGM0068 - TB2 2024/25

31 of 31

End of Term Unit Evaluation

https://go.blueja.io/v6_MavUug02kfifVQDUC-Q

GEOGM0068 - TB2 2024/25