SlideShare ist ein Scribd-Unternehmen logo
1 von 40
Downloaden Sie, um offline zu lesen
Guenther Goerz, Chiara Seidl, Martin Thiering
Bibliotheca Hertziana –Max Planck Institute for the History of Art, Rome
FAU Erlangen-Nuremberg, Department of Computer Science, Digital
Humanities
Technical University Berlin, Department of Linguistics
Linked Biondo
Modelling Geographical Features
in Renaissance Texts and Maps
G.	Goerz,	FAU,	CS	DH	&	BHR	 2
Research Question and Methodology	
•  Cognition of geographical space in history:
How would you start ?
–  Toponyms and definite place descriptions
•  Text annotation: Named Entity Recognition,
geographic verification
–  Spatial relations: topology, orientation, distance,…
•  Cognitive-linguistic annotation
Constructions figure- spatial_indicator – ground
–  Comparisons with contemporary maps
•  Cognitive maps (Common Sense Geography)
–  Spatial objects and relations
–  Epistemological modelling
G.	Goerz,	FAU,	CS	DH	&	BHR	 3
Problem statement
•  Common sense conceptualizations of geographic
concepts and relations in ancient and early modern
texts and maps
–  Analytic methods of cognitive computational linguistics
Corpus construction, annotation, and parsing
–  Formal two-level representation
•  (Cognitive) linguistic
•  Conceptual – general semantics (onto-logical)
–  Linguistic and historical interpretation and evaluation
–  Long term goal: Synthetic
Reconstruction of cognitive maps/sketches
G.	Goerz,	FAU,	CS	DH	&	BHR	 4
Sources: Preparation and Conditioning	
•  Flavio Biondo Italia Illustrata (1474)
–  “Topographically ordered historical account” (Clavuot)
–  Connecting antiquity and presence
•  Text editions: Pontari (la), White, Castner (la, en)
•  Preprocessing
–  OCR correction, sentence separation, etc.
–  Word lists, statistics, n-grams
•  scripts, antconc, voyant-tools, ...
–  Concordance (KWIC)
–  Part-of-Speech Tagging (TreeTagger, Collatinus)
–  wordij network: “semantic” clustering
•  Map image processing
G.	Goerz,	FAU,	CS	DH	&	BHR	 5
Toponyms and spatial relations
•  Text annotation (semi-automatic)
– Recogito 2 annotation tool with geographic
verfication (gazetteers, e.g. Pleiades)
– Spatial Role Labeling (manually: brat)
•  Machine Learning? … corpus problem
– Narrative order: virtual trips
•  Map annotation
– Recogito 2 (incl. visualization)
– Cartometric investigations (Guckelsberger)
G.	Goerz,	FAU,	CS	DH	&	BHR	 6
Investigations on Renaissance maps
•  Biondo’s use of maps ? (“Petrarca”, Ptolemy)
•  Which maps to choose ?
– Six large maps of Italy (15. c.; Milanesi
2007/08); Paulinus Minorita (14. c.); Tabula
Peutingeriana
– Ptolemaic maps (ca. 30 traditional and
“novae” – mostly after 1450!)
– Sea charts (Portolans) before 1450 (max. 10)
G.	Goerz,	FAU,	CS	DH	&	BHR	 7
G.	Goerz,	FAU,	CS	DH	&	BHR	 8
G.	Goerz,	FAU,	CS	DH	&	BHR	 9
L23nova	
G.	Goerz,	FAU,	CS	DH	&	BHR	 10
DARE map view of Ptolemaeus L23n
G.	Goerz,	FAU,	CS	DH	&	BHR	 11
Generated research data
•  Recogito data export
– CSV (tables), GeoJSON, RDF/OA, JSON-LD,…
– Case study: Ptolemaic tabulae novae & text
•  Spatial Role Labeling
G.	Goerz,	FAU,	CS	DH	&	BHR	 12	
according to
predefined
cognitive-linguistic
taxonomy
Table for Ptolemy L23nova
G.	Goerz,	FAU,	CS	DH	&	BHR	 13
Cognitive linguistics and Gestalt theory
•  Mental models are based on universal cognitive
mechanisms and Gestalt-theoretical principles (!)
– cf. “Primary theory” (Smith/Mark 2001)
•  Particularly relevant: spatial division of figure and
background in identifying and locating objects
•  Spatial relations are represented through grammatical
markers and semantic fields
•  From representational viewpoint: Mental models store
information on events and objects of the external
world, especially
–  for orientation in space and references to places,
for topological and geometrical knowledge
G.	Goerz,	FAU,	C.S.	-	AG	DH
Spatial Construals and its Spatial
Parameters
•  Gestalt principles of figure–ground (TRAJECTOR-
LANDMARK) asymmetries;
trajectory/path of TR and LM
•  OBJECT CLASSIFICATIONS, mental rotations, 2,5/3-D
sketch, geometrical dimensions
•  FRAMES OF REFERENCE (relative; intrinsic; absolute)
•  TOPONYMS
(place / city names, buildings, bridges, churches,
fountains, walls, streets, squares, rivers, hills, gates,
memorials, temples, sites, regions, etc.)
	
G.	Goerz,	FAU,	C.S.	-	AG	DH
Spatial Construals and its Spatial
Parameters
	
•  LANDMARKS
•  DISTANCES (scale, scope, size),
encoded in adjectives, adverbs, verbs but mostly in
adpositions and case systems
•  METRICAL SYSTEMS
(verbal systems such as posture verbs, classificatory
verbs and case systems)
•  PERSPECTIVE
(bird’s eye, hodological, vectorial perspective)
•  ELEMENTS OF COMMON SENSE KNOWLEDGE
(traveller reports, myths etc.)
	
G.	Goerz,	FAU,	C.S.	-	AG	DH
Spatial Construals and its Spatial
Parameters
•  MOTION EVENT: SOURCE = Point of departure of TR
•  PATH/TRAJECTORY = Movement of TR from SOURCE[TR(X)]
to GOAL[LM(Y)]
•  GOAL = GOAL of TR'S movement to LM(Y);
often a container such as a room, city, town, church
etc.
•  DISTANCE = proximate1[PROX], medial2[MED], distal3[DIST]
between TR and LM
•  PROFILE = TRAJECTOR'S specification of LANDMARK
•  Conceptualization of spatial structure: Static concepts
include a REGION, LOCATION, and dynamic concepts
include PATH and PLACEMENT of TR
G.	Goerz,	FAU,	C.S.	-	AG	DH
Spatial Role Labeling
in a Cognitive Linguistic Framework
•  Given: SpatialML annotation scheme
•  Definition of a brat “configuration” (taxonomy)
•  Annotation sentence by sentence with brat, manually
•  Parallel text: transfer to Latin
•  XML/RDF export
–  to be combined with dependency relations
–  information integration with NER results
•  Evaluation and Interpretation
–  Evaluation of the use of prototypical functions with
lemmata (Latin/English) in order of frequency
–  Specially: Landmarks, toponyms, frames of reference,
perspectives
G.	Goerz,	FAU,	C.S.	-	AG	DH
Part-of-
Speech
Analysis
(excerpt)
G.	Goerz,	FAU,	CS	DH	&	BHR	 21	
T1 	region	6	14 	Etruriae	
T2 	definite_description	0	14 	Finis	Etruriae	
T3 	river	18	25 	Tiberim	
T4 	settlement	40	45 	Romam	
T6 	spatial_indicator	15	17 	ad	
E1 	spatial_indicator:T6	figure:T1	ground:T3	
T7 	action	30	39 	perducens	
						T1	/	T2																													T6																						T3														T7												T4	
© Kordjamshidi 2013
Machine Learning for Spatial Role
Labeling?
•  Kordjamshidi et al. (ca. 2010 ff.), KU Leuven
–  Hybrid approach (with klog rules; cf. “explainable AI”)
–  CoNLL competition, training corpus (campus)
–  new hybrid framework since 2015 (Univ. Illinois)
•  Problem of training data
–  only for English, different text sort
–  Latin from manual labeling, very small
•  Experiment with Sarah Schulz (U. Stuttgart, 2018)
–  Software (above) incomplete: missing modules
–  Therefore using open software: mod:nlpnet (base:
NLTK, numpy)
Machine Learning for Spatial Role
Labeling?
•  Experiment with Sarah Schulz : mod:nlpnet
–  Multilayer perceptron for POS tagging
–  Convolutional NN for SRL tagging
•  adapted to SpRL, actually figure–sp_ind-ground
–  Results for English with problematic training
corpus: anything but representative
•  find identifier: accuracy 0.96
•  find figure and ground: precision 0.54, recall 0.25
•  recall figure: 0.19, ground: 0.30
wordij network: “semantic” clustering
G.	Goerz,	FAU,	CS	DH	&	BHR	 24
G.	Goerz,	FAU,	CS	DH	&	BHR	 25
Semantic enhancement of data	
•  What is the meaning of annotations ?
•  Semantics of annotation components defined in terms of a
formal ontology
•  Formal ontology: knowledge modelling
–  two methodological levels
•  Reference ontology CIDOC CRM (ISO 21127) with extension
CRMgeo
–  CRM event-based; linguistic-pragmatic approach
–  Refining of domain object descriptions with technical
terms (“types”) from thesauri (“Pleiades vocabulary”)
–  Use of authority files	
G.	Goerz,	FAU,	CS	DH	&	BHR	 26
participate in
Actors Conceptual Objects
Physical Entities
Temporal Entities
affect
Types
refine
Appellations
identify/name
location
occur atwithin
Time-Spans
Places
CIDOC CRM
Top Level Classes
©	T.	Gill	 G.	Goerz,	FAU,	CS	DH	&	BHR	 27
CRM + CRMgeo in OWL-DL (Protégé)
G.	Goerz,	FAU,	CS	DH	&	BHR	 28
CRM + CRMgeo in OWL-DL (Protégé)	
G.	Goerz,	FAU,	CS	DH	&	BHR	 29
hmap + CRM + CRMgeo
•  hmap: domain ontology for historical maps
–  Map Metadata
ID, Cartographer, Creator, Title, Place, Time,
Size, Material, Technique, Projection, Scale,
Orientation, Reference System, …
–  Image (Reproduction) Metadata: Map, URL,
Dimension, Rights,...
–  Content: Annotated places and connections
UUID, Transcription, Anchor, Type, URI
(Gazetteer), Label, Lat, Lng, Place Type,
Verif_status, …
G.	Goerz,	FAU,	CS	DH	&	BHR	 30
In search of a semantic platform
•  How can we perform the semantic
enhancement of annotation data ?
•  How can we publish them as Linked
Open Data ?
•  Transformation with VRE WissKI,
usage as Linked Open Data platform
•  SPARQL query interface, RDF export,…
G.	Goerz,	FAU,	CS	DH	&	BHR	 31
What is WissKI ?
(“Scholarly Communication Infrastructure“)
•  Extension of CMS Drupal, customizable
•  Web-based multi-user system
•  Open source & open standards (Semantic
Web)
•  Object-based documentation, multiple media
types
•  Ontology-based representation (ECRM/OWL),
extensible by application ontologies &
controlled vocabularies
G.	Goerz,	FAU,	CS	DH	&	BHR	 32
WissKI Software Architecture
G.	Goerz,	FAU,	CS	DH	&	BHR	 33
Layered	
Ontologies	
34	G.	Goerz,	FAU,	CS	DH	&	BHR
WissKI
•  Create, Navigate, Find : main modes
•  Create: Data input
– Form based or text based
– (automatic) linking
– data enrichment with external sources
– various import and export formats
•  Pathbuilder: define semantics of fields
G.	Goerz,	FAU,	CS	DH	&	BHR	 35
Input Form: Semantic Modelling
36	
Albrecht Dürer
Nürnberg
E84 Information Carrier
→ P108i was produced by →
E12 Production
→ P14 carried out by →
E21 Person
→ P131 is identified by →
E82 Actor Appellation
→ P3 has note →
„Albrecht Dürer“
E84 Information Carrier
→ P108i was produced by →
E12 Production
→ P7 took place at →
E53 Place
→ P87 is identified by →
E48 Place Name
→ P3 has note →
„Nürnberg “
©	HohmannG.	Goerz,	FAU,	CS	DH	&	BHR
Linked Open Data (... SPARQL endpoint)
DNB	GND	
Geonames	
BritishMuseum	
Linking	Open	Data	cloud	diagram	2017,	by	Andrejs	Abele,	John	P.	McCrae,	Paul	Buitelaar,	Anja	
Jentzsch	and	Richard	Cyganiak.	http://lod-cloud.net/	
G.	Goerz,	FAU,	CS	DH	 37
Next steps
•  “In the last analysis all maps are cognitive maps”
(Blakemore & Harley)
•  Abstract conceptual knowledge represented by
object schemata – propositional and “depictional”
•  Translation of descriptions of spatial objects and their
spatial relations
–  Triples [figure, spatial_indicator, ground]
–  Constructing a spatial property graph
–  Generating sketches of cognitive maps to represent and
process reifications of cognitive objects on an
epistemological level, i.e. frame of reference, topology,
direction, trajectory, distance, and shape
•  Spatial reasoning ?
G.	Goerz,	FAU,	C.S.	-	AG	DH
Grazie!
G.	Goerz,	FAU,	CS	DH	&	BHR	 39
G.	Goerz,	FAU,	CS	DH	&	BHR	 40	
CRM + CRMgeo
in OWL-DL
(Protégé)

Weitere ähnliche Inhalte

Mehr von Erlangen Artificial Intelligence & Machine Learning Meetup (7)

NLP@DATEV: Setting up a domain specific language model, Dr. Jonas Rende & Tho...
NLP@DATEV: Setting up a domain specific language model, Dr. Jonas Rende & Tho...NLP@DATEV: Setting up a domain specific language model, Dr. Jonas Rende & Tho...
NLP@DATEV: Setting up a domain specific language model, Dr. Jonas Rende & Tho...
 
Knowledge Graphs, Daria Stepanova, Bosch Center for Artificial Intelligence
Knowledge Graphs, Daria Stepanova, Bosch Center for Artificial IntelligenceKnowledge Graphs, Daria Stepanova, Bosch Center for Artificial Intelligence
Knowledge Graphs, Daria Stepanova, Bosch Center for Artificial Intelligence
 
AI applications in education, Pascal Zoleko, Flexudy
AI applications in education, Pascal Zoleko, FlexudyAI applications in education, Pascal Zoleko, Flexudy
AI applications in education, Pascal Zoleko, Flexudy
 
Learning global pooling operators in deep neural networks for image retrieval...
Learning global pooling operators in deep neural networks for image retrieval...Learning global pooling operators in deep neural networks for image retrieval...
Learning global pooling operators in deep neural networks for image retrieval...
 
XGBoostLSS - An extension of XGBoost to probabilistic forecasting, Alexander ...
XGBoostLSS - An extension of XGBoost to probabilistic forecasting, Alexander ...XGBoostLSS - An extension of XGBoost to probabilistic forecasting, Alexander ...
XGBoostLSS - An extension of XGBoost to probabilistic forecasting, Alexander ...
 
Machine Learning Operations & Azure
Machine Learning Operations & AzureMachine Learning Operations & Azure
Machine Learning Operations & Azure
 
Best practices for structuring Machine Learning code
Best practices for structuring Machine Learning codeBest practices for structuring Machine Learning code
Best practices for structuring Machine Learning code
 

Kürzlich hochgeladen

Infrared simulation and processing on Nvidia platforms
Infrared simulation and processing on Nvidia platformsInfrared simulation and processing on Nvidia platforms
Infrared simulation and processing on Nvidia platformsYoss Cohen
 
Genislab builds better products and faster go-to-market with Lean project man...
Genislab builds better products and faster go-to-market with Lean project man...Genislab builds better products and faster go-to-market with Lean project man...
Genislab builds better products and faster go-to-market with Lean project man...Farhan Tariq
 
Transcript: New from BookNet Canada for 2024: BNC SalesData and LibraryData -...
Transcript: New from BookNet Canada for 2024: BNC SalesData and LibraryData -...Transcript: New from BookNet Canada for 2024: BNC SalesData and LibraryData -...
Transcript: New from BookNet Canada for 2024: BNC SalesData and LibraryData -...BookNet Canada
 
Accelerating Enterprise Software Engineering with Platformless
Accelerating Enterprise Software Engineering with PlatformlessAccelerating Enterprise Software Engineering with Platformless
Accelerating Enterprise Software Engineering with PlatformlessWSO2
 
4. Cobus Valentine- Cybersecurity Threats and Solutions for the Public Sector
4. Cobus Valentine- Cybersecurity Threats and Solutions for the Public Sector4. Cobus Valentine- Cybersecurity Threats and Solutions for the Public Sector
4. Cobus Valentine- Cybersecurity Threats and Solutions for the Public Sectoritnewsafrica
 
Time Series Foundation Models - current state and future directions
Time Series Foundation Models - current state and future directionsTime Series Foundation Models - current state and future directions
Time Series Foundation Models - current state and future directionsNathaniel Shimoni
 
Landscape Catalogue 2024 Australia-1.pdf
Landscape Catalogue 2024 Australia-1.pdfLandscape Catalogue 2024 Australia-1.pdf
Landscape Catalogue 2024 Australia-1.pdfAarwolf Industries LLC
 
Generative Artificial Intelligence: How generative AI works.pdf
Generative Artificial Intelligence: How generative AI works.pdfGenerative Artificial Intelligence: How generative AI works.pdf
Generative Artificial Intelligence: How generative AI works.pdfIngrid Airi González
 
A Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersA Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersNicole Novielli
 
A Framework for Development in the AI Age
A Framework for Development in the AI AgeA Framework for Development in the AI Age
A Framework for Development in the AI AgeCprime
 
2024 April Patch Tuesday
2024 April Patch Tuesday2024 April Patch Tuesday
2024 April Patch TuesdayIvanti
 
Connecting the Dots for Information Discovery.pdf
Connecting the Dots for Information Discovery.pdfConnecting the Dots for Information Discovery.pdf
Connecting the Dots for Information Discovery.pdfNeo4j
 
Decarbonising Buildings: Making a net-zero built environment a reality
Decarbonising Buildings: Making a net-zero built environment a realityDecarbonising Buildings: Making a net-zero built environment a reality
Decarbonising Buildings: Making a net-zero built environment a realityIES VE
 
Top 10 Hubspot Development Companies in 2024
Top 10 Hubspot Development Companies in 2024Top 10 Hubspot Development Companies in 2024
Top 10 Hubspot Development Companies in 2024TopCSSGallery
 
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...Wes McKinney
 
React JS; all concepts. Contains React Features, JSX, functional & Class comp...
React JS; all concepts. Contains React Features, JSX, functional & Class comp...React JS; all concepts. Contains React Features, JSX, functional & Class comp...
React JS; all concepts. Contains React Features, JSX, functional & Class comp...Karmanjay Verma
 
UiPath Community: Communication Mining from Zero to Hero
UiPath Community: Communication Mining from Zero to HeroUiPath Community: Communication Mining from Zero to Hero
UiPath Community: Communication Mining from Zero to HeroUiPathCommunity
 
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Alkin Tezuysal
 
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...panagenda
 
Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...
Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...
Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...Nikki Chapple
 

Kürzlich hochgeladen (20)

Infrared simulation and processing on Nvidia platforms
Infrared simulation and processing on Nvidia platformsInfrared simulation and processing on Nvidia platforms
Infrared simulation and processing on Nvidia platforms
 
Genislab builds better products and faster go-to-market with Lean project man...
Genislab builds better products and faster go-to-market with Lean project man...Genislab builds better products and faster go-to-market with Lean project man...
Genislab builds better products and faster go-to-market with Lean project man...
 
Transcript: New from BookNet Canada for 2024: BNC SalesData and LibraryData -...
Transcript: New from BookNet Canada for 2024: BNC SalesData and LibraryData -...Transcript: New from BookNet Canada for 2024: BNC SalesData and LibraryData -...
Transcript: New from BookNet Canada for 2024: BNC SalesData and LibraryData -...
 
Accelerating Enterprise Software Engineering with Platformless
Accelerating Enterprise Software Engineering with PlatformlessAccelerating Enterprise Software Engineering with Platformless
Accelerating Enterprise Software Engineering with Platformless
 
4. Cobus Valentine- Cybersecurity Threats and Solutions for the Public Sector
4. Cobus Valentine- Cybersecurity Threats and Solutions for the Public Sector4. Cobus Valentine- Cybersecurity Threats and Solutions for the Public Sector
4. Cobus Valentine- Cybersecurity Threats and Solutions for the Public Sector
 
Time Series Foundation Models - current state and future directions
Time Series Foundation Models - current state and future directionsTime Series Foundation Models - current state and future directions
Time Series Foundation Models - current state and future directions
 
Landscape Catalogue 2024 Australia-1.pdf
Landscape Catalogue 2024 Australia-1.pdfLandscape Catalogue 2024 Australia-1.pdf
Landscape Catalogue 2024 Australia-1.pdf
 
Generative Artificial Intelligence: How generative AI works.pdf
Generative Artificial Intelligence: How generative AI works.pdfGenerative Artificial Intelligence: How generative AI works.pdf
Generative Artificial Intelligence: How generative AI works.pdf
 
A Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersA Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software Developers
 
A Framework for Development in the AI Age
A Framework for Development in the AI AgeA Framework for Development in the AI Age
A Framework for Development in the AI Age
 
2024 April Patch Tuesday
2024 April Patch Tuesday2024 April Patch Tuesday
2024 April Patch Tuesday
 
Connecting the Dots for Information Discovery.pdf
Connecting the Dots for Information Discovery.pdfConnecting the Dots for Information Discovery.pdf
Connecting the Dots for Information Discovery.pdf
 
Decarbonising Buildings: Making a net-zero built environment a reality
Decarbonising Buildings: Making a net-zero built environment a realityDecarbonising Buildings: Making a net-zero built environment a reality
Decarbonising Buildings: Making a net-zero built environment a reality
 
Top 10 Hubspot Development Companies in 2024
Top 10 Hubspot Development Companies in 2024Top 10 Hubspot Development Companies in 2024
Top 10 Hubspot Development Companies in 2024
 
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
 
React JS; all concepts. Contains React Features, JSX, functional & Class comp...
React JS; all concepts. Contains React Features, JSX, functional & Class comp...React JS; all concepts. Contains React Features, JSX, functional & Class comp...
React JS; all concepts. Contains React Features, JSX, functional & Class comp...
 
UiPath Community: Communication Mining from Zero to Hero
UiPath Community: Communication Mining from Zero to HeroUiPath Community: Communication Mining from Zero to Hero
UiPath Community: Communication Mining from Zero to Hero
 
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
 
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
 
Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...
Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...
Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...
 

Ai for cultural history

  • 1. Guenther Goerz, Chiara Seidl, Martin Thiering Bibliotheca Hertziana –Max Planck Institute for the History of Art, Rome FAU Erlangen-Nuremberg, Department of Computer Science, Digital Humanities Technical University Berlin, Department of Linguistics Linked Biondo Modelling Geographical Features in Renaissance Texts and Maps
  • 3. Research Question and Methodology •  Cognition of geographical space in history: How would you start ? –  Toponyms and definite place descriptions •  Text annotation: Named Entity Recognition, geographic verification –  Spatial relations: topology, orientation, distance,… •  Cognitive-linguistic annotation Constructions figure- spatial_indicator – ground –  Comparisons with contemporary maps •  Cognitive maps (Common Sense Geography) –  Spatial objects and relations –  Epistemological modelling G. Goerz, FAU, CS DH & BHR 3
  • 4. Problem statement •  Common sense conceptualizations of geographic concepts and relations in ancient and early modern texts and maps –  Analytic methods of cognitive computational linguistics Corpus construction, annotation, and parsing –  Formal two-level representation •  (Cognitive) linguistic •  Conceptual – general semantics (onto-logical) –  Linguistic and historical interpretation and evaluation –  Long term goal: Synthetic Reconstruction of cognitive maps/sketches G. Goerz, FAU, CS DH & BHR 4
  • 5. Sources: Preparation and Conditioning •  Flavio Biondo Italia Illustrata (1474) –  “Topographically ordered historical account” (Clavuot) –  Connecting antiquity and presence •  Text editions: Pontari (la), White, Castner (la, en) •  Preprocessing –  OCR correction, sentence separation, etc. –  Word lists, statistics, n-grams •  scripts, antconc, voyant-tools, ... –  Concordance (KWIC) –  Part-of-Speech Tagging (TreeTagger, Collatinus) –  wordij network: “semantic” clustering •  Map image processing G. Goerz, FAU, CS DH & BHR 5
  • 6. Toponyms and spatial relations •  Text annotation (semi-automatic) – Recogito 2 annotation tool with geographic verfication (gazetteers, e.g. Pleiades) – Spatial Role Labeling (manually: brat) •  Machine Learning? … corpus problem – Narrative order: virtual trips •  Map annotation – Recogito 2 (incl. visualization) – Cartometric investigations (Guckelsberger) G. Goerz, FAU, CS DH & BHR 6
  • 7. Investigations on Renaissance maps •  Biondo’s use of maps ? (“Petrarca”, Ptolemy) •  Which maps to choose ? – Six large maps of Italy (15. c.; Milanesi 2007/08); Paulinus Minorita (14. c.); Tabula Peutingeriana – Ptolemaic maps (ca. 30 traditional and “novae” – mostly after 1450!) – Sea charts (Portolans) before 1450 (max. 10) G. Goerz, FAU, CS DH & BHR 7
  • 11. DARE map view of Ptolemaeus L23n G. Goerz, FAU, CS DH & BHR 11
  • 12. Generated research data •  Recogito data export – CSV (tables), GeoJSON, RDF/OA, JSON-LD,… – Case study: Ptolemaic tabulae novae & text •  Spatial Role Labeling G. Goerz, FAU, CS DH & BHR 12 according to predefined cognitive-linguistic taxonomy
  • 13. Table for Ptolemy L23nova G. Goerz, FAU, CS DH & BHR 13
  • 14. Cognitive linguistics and Gestalt theory •  Mental models are based on universal cognitive mechanisms and Gestalt-theoretical principles (!) – cf. “Primary theory” (Smith/Mark 2001) •  Particularly relevant: spatial division of figure and background in identifying and locating objects •  Spatial relations are represented through grammatical markers and semantic fields •  From representational viewpoint: Mental models store information on events and objects of the external world, especially –  for orientation in space and references to places, for topological and geometrical knowledge G. Goerz, FAU, C.S. - AG DH
  • 15. Spatial Construals and its Spatial Parameters •  Gestalt principles of figure–ground (TRAJECTOR- LANDMARK) asymmetries; trajectory/path of TR and LM •  OBJECT CLASSIFICATIONS, mental rotations, 2,5/3-D sketch, geometrical dimensions •  FRAMES OF REFERENCE (relative; intrinsic; absolute) •  TOPONYMS (place / city names, buildings, bridges, churches, fountains, walls, streets, squares, rivers, hills, gates, memorials, temples, sites, regions, etc.) G. Goerz, FAU, C.S. - AG DH
  • 16. Spatial Construals and its Spatial Parameters •  LANDMARKS •  DISTANCES (scale, scope, size), encoded in adjectives, adverbs, verbs but mostly in adpositions and case systems •  METRICAL SYSTEMS (verbal systems such as posture verbs, classificatory verbs and case systems) •  PERSPECTIVE (bird’s eye, hodological, vectorial perspective) •  ELEMENTS OF COMMON SENSE KNOWLEDGE (traveller reports, myths etc.) G. Goerz, FAU, C.S. - AG DH
  • 17. Spatial Construals and its Spatial Parameters •  MOTION EVENT: SOURCE = Point of departure of TR •  PATH/TRAJECTORY = Movement of TR from SOURCE[TR(X)] to GOAL[LM(Y)] •  GOAL = GOAL of TR'S movement to LM(Y); often a container such as a room, city, town, church etc. •  DISTANCE = proximate1[PROX], medial2[MED], distal3[DIST] between TR and LM •  PROFILE = TRAJECTOR'S specification of LANDMARK •  Conceptualization of spatial structure: Static concepts include a REGION, LOCATION, and dynamic concepts include PATH and PLACEMENT of TR G. Goerz, FAU, C.S. - AG DH
  • 18. Spatial Role Labeling in a Cognitive Linguistic Framework •  Given: SpatialML annotation scheme •  Definition of a brat “configuration” (taxonomy) •  Annotation sentence by sentence with brat, manually •  Parallel text: transfer to Latin •  XML/RDF export –  to be combined with dependency relations –  information integration with NER results •  Evaluation and Interpretation –  Evaluation of the use of prototypical functions with lemmata (Latin/English) in order of frequency –  Specially: Landmarks, toponyms, frames of reference, perspectives G. Goerz, FAU, C.S. - AG DH
  • 20.
  • 21. G. Goerz, FAU, CS DH & BHR 21 T1 region 6 14 Etruriae T2 definite_description 0 14 Finis Etruriae T3 river 18 25 Tiberim T4 settlement 40 45 Romam T6 spatial_indicator 15 17 ad E1 spatial_indicator:T6 figure:T1 ground:T3 T7 action 30 39 perducens T1 / T2 T6 T3 T7 T4 © Kordjamshidi 2013
  • 22. Machine Learning for Spatial Role Labeling? •  Kordjamshidi et al. (ca. 2010 ff.), KU Leuven –  Hybrid approach (with klog rules; cf. “explainable AI”) –  CoNLL competition, training corpus (campus) –  new hybrid framework since 2015 (Univ. Illinois) •  Problem of training data –  only for English, different text sort –  Latin from manual labeling, very small •  Experiment with Sarah Schulz (U. Stuttgart, 2018) –  Software (above) incomplete: missing modules –  Therefore using open software: mod:nlpnet (base: NLTK, numpy)
  • 23. Machine Learning for Spatial Role Labeling? •  Experiment with Sarah Schulz : mod:nlpnet –  Multilayer perceptron for POS tagging –  Convolutional NN for SRL tagging •  adapted to SpRL, actually figure–sp_ind-ground –  Results for English with problematic training corpus: anything but representative •  find identifier: accuracy 0.96 •  find figure and ground: precision 0.54, recall 0.25 •  recall figure: 0.19, ground: 0.30
  • 24. wordij network: “semantic” clustering G. Goerz, FAU, CS DH & BHR 24
  • 26. Semantic enhancement of data •  What is the meaning of annotations ? •  Semantics of annotation components defined in terms of a formal ontology •  Formal ontology: knowledge modelling –  two methodological levels •  Reference ontology CIDOC CRM (ISO 21127) with extension CRMgeo –  CRM event-based; linguistic-pragmatic approach –  Refining of domain object descriptions with technical terms (“types”) from thesauri (“Pleiades vocabulary”) –  Use of authority files G. Goerz, FAU, CS DH & BHR 26
  • 27. participate in Actors Conceptual Objects Physical Entities Temporal Entities affect Types refine Appellations identify/name location occur atwithin Time-Spans Places CIDOC CRM Top Level Classes © T. Gill G. Goerz, FAU, CS DH & BHR 27
  • 28. CRM + CRMgeo in OWL-DL (Protégé) G. Goerz, FAU, CS DH & BHR 28
  • 29. CRM + CRMgeo in OWL-DL (Protégé) G. Goerz, FAU, CS DH & BHR 29
  • 30. hmap + CRM + CRMgeo •  hmap: domain ontology for historical maps –  Map Metadata ID, Cartographer, Creator, Title, Place, Time, Size, Material, Technique, Projection, Scale, Orientation, Reference System, … –  Image (Reproduction) Metadata: Map, URL, Dimension, Rights,... –  Content: Annotated places and connections UUID, Transcription, Anchor, Type, URI (Gazetteer), Label, Lat, Lng, Place Type, Verif_status, … G. Goerz, FAU, CS DH & BHR 30
  • 31. In search of a semantic platform •  How can we perform the semantic enhancement of annotation data ? •  How can we publish them as Linked Open Data ? •  Transformation with VRE WissKI, usage as Linked Open Data platform •  SPARQL query interface, RDF export,… G. Goerz, FAU, CS DH & BHR 31
  • 32. What is WissKI ? (“Scholarly Communication Infrastructure“) •  Extension of CMS Drupal, customizable •  Web-based multi-user system •  Open source & open standards (Semantic Web) •  Object-based documentation, multiple media types •  Ontology-based representation (ECRM/OWL), extensible by application ontologies & controlled vocabularies G. Goerz, FAU, CS DH & BHR 32
  • 35. WissKI •  Create, Navigate, Find : main modes •  Create: Data input – Form based or text based – (automatic) linking – data enrichment with external sources – various import and export formats •  Pathbuilder: define semantics of fields G. Goerz, FAU, CS DH & BHR 35
  • 36. Input Form: Semantic Modelling 36 Albrecht Dürer Nürnberg E84 Information Carrier → P108i was produced by → E12 Production → P14 carried out by → E21 Person → P131 is identified by → E82 Actor Appellation → P3 has note → „Albrecht Dürer“ E84 Information Carrier → P108i was produced by → E12 Production → P7 took place at → E53 Place → P87 is identified by → E48 Place Name → P3 has note → „Nürnberg “ © HohmannG. Goerz, FAU, CS DH & BHR
  • 37. Linked Open Data (... SPARQL endpoint) DNB GND Geonames BritishMuseum Linking Open Data cloud diagram 2017, by Andrejs Abele, John P. McCrae, Paul Buitelaar, Anja Jentzsch and Richard Cyganiak. http://lod-cloud.net/ G. Goerz, FAU, CS DH 37
  • 38. Next steps •  “In the last analysis all maps are cognitive maps” (Blakemore & Harley) •  Abstract conceptual knowledge represented by object schemata – propositional and “depictional” •  Translation of descriptions of spatial objects and their spatial relations –  Triples [figure, spatial_indicator, ground] –  Constructing a spatial property graph –  Generating sketches of cognitive maps to represent and process reifications of cognitive objects on an epistemological level, i.e. frame of reference, topology, direction, trajectory, distance, and shape •  Spatial reasoning ? G. Goerz, FAU, C.S. - AG DH
  • 40. G. Goerz, FAU, CS DH & BHR 40 CRM + CRMgeo in OWL-DL (Protégé)