artificial intelligence aided geospatial knowledge databases. … · 2020. 2. 3. · artificial...

28
Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations Adam Iwaniak Wroclaw Institute of Spatial Information and Artificial Intelligence “Copernicus and Artificial Intelligence” workshop 28 January 2020 Brussels, Belgium

Upload: others

Post on 23-Aug-2020

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

Artificial

Intelligence aided

Geospatial Knowledge

Databases. Creodias

and other

implementations

Adam Iwaniak Wroclaw Institute of Spatial

Information and Artificial

Intelligence

“Copernicus and Artificial Intelligence” workshop

28 January 2020

Brussels, Belgium

Page 2: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

Linked Data

Data on the Web

Spatial data on the Web

Geospatial industry SDI

Integration of WWW and GIS

Internet industry

Page 3: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

RDF model

Building 1461215_1.0024.26/1.1_BUD

is located on the parcel no. 141215_1.0024.26/1

<http://pzgik.geoportal.gov.pl/egib/1461215_1.0024.26.1_BUD>

<is_contained_in>

<http://mapy.geoportal.gov.pl/imap/?identifyParcel=141215_1.0024.26

/1>

1461215_1.0024.26.1_B

UD

141215_1.0024.26/1

WR1K/00242114/4

*LMR – land and mortgage register record

Page 4: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

Belvedere Palace - a real

objecthttps://pzgik.geoportal.gov.pl/e

gib/146510_8.0615.11.1_BUD

Land and building

records

https://pzgik.geoportal.

gov.pl/egib/146510_8.061

5.11.1_BUD

BDOT10k

https://pzgik.geoportal.go

v.pl/bdot10k/building/Belw

eder

Orthophotomap

https://pzgik.geoportal.go

v.pl/orto/dataset/8830040

/

NID

https://zabytek.pl/zasoby/Zabyt

ekNieruchomy/PL.1.9.ZIPOZ.NID_N

_14_PH.8443

OpenStreetMap

https://www.openstreetmap.org/w

ay/25748432

WikiData

https://www.wikidata.org/

wiki/Q645474

DBpedia

http://dbpedia.org/page/Belwe

der

MPZP

https://wrotamazowsza.pl/m

pzp/121516-288589-251115

Website

http://www.polskiezabytki.pl

/m/obiekt/7837/Warszawa_-

_Palac_Belweder/

Page 5: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

Thematic data

State Geodetic and Cartographic

Resource

Localgoverments

WWW

BDOT10k

Orthoimagery

Land and building records

MPZP

DBpedia

WikiData

Websites

OpenStreetMap

NID

Page 6: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

Knowledge graph

Page 7: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

Spatial data to RDF conversion

Dxf, GML

files

Spatial Analysis

RDF Repository

AllegroGraph

GeoMedia

SPARQL

RDF

files

Import

/export

SQLWIZIPISI

Semantic

Components

Import

/export

DB

Graph Analysis

Questions:

1. Find parcels of Adam Iwaniak family

2. Find property of Adam Iwaniak family

Page 8: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

SNA analysis - centrality measures

• How many direct connections does a

node have with other nodes in the

network?

• How important is a given node in the

network / group?

• Based on the number of connections

for a given vertex.

1

0,14

0,14

0,14

0,14

0,14

0,28

0,28

Page 9: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

INSPIRE metadata Catalogue and WWW

integration

Innovativeness of the new method of publishing

data, metadata and spatial data services:

• Metadata is available for search engines

• Metadata representation in HTML + semantic

annotations

• Enrichment of metadata with links to other

resources: GEMET, GeoNames, DBPedia.

Page 10: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

RDFHTML+RDFa

Spatial Data Base

SOoI

Geoportal

GoogleIndexing

INSPIRE data in HTML + RDFa

SOoI – Spatial Object of Interest /point, line and area/

Page 11: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020A.Iwaniak, The INSPIRE Conference 2014, Aalborg, 16-20 June 2014

GeoDCAT-AP profile Compliance

Page 12: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

act

Data and Information Access Services –

Copernicus DIAS, 2018

Page 13: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

DIAS - Data and Information Access

Services

• A platform for quick access to a huge resource of Earth Observation data (satellite imagery)

and Copernicus themed services

• Easy and quick search, processing and sharing

of satellite data

• Quick access to satellite imagery via virtual machines

Page 14: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

Data catalogue

browser.creodias.eu

OGC CSW

Page 15: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

SPARQL endpoint (graph database /rdf repository)

sparql.creodias.eu/…

Page 16: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

CREODIAS Knowledge Database –

World divided into hexagons

Average Hexagon Area (km2)

Average Hexagon Edge Length (km)

Number of objects

1,770.3235517 22.606379400 288,122

Uber’s Hexagonal Hierarchical Spatial Index

Page 17: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

ids, geom

Spatial objectsIds, geom

Attribute Value

TitleS2A_MSIL1C_20180412T083651_N0206_R064_T35RPH_20180412T161307.SAFE

Organisation ESA

Start Date 2018-04-12 08:36:51

Completion Date 2018-04-12 08:36:51

Mission Sentinel-2

Cloud Coverage 0.01

Product Type L1C

Instrument MSI

Processing Level LEVELL1C

Processing Start 2018-04-12 16:13:07

Resource Size 755592104.0

Platform S2A

Product Identifier

/eodata/Sentinel-2/MSI/L1C/2018/04/12/S2A_MSIL1C_20180412T083651_N0206_R064_T35RPH_20180412T161307.SAFE

Orbit Number 14647

Orbit Direction descending

NSSDC Identifier 2015-000A

INSPIRE metadata ID740a2303-b99d-4f0e-b914-18a679c4f827

SEMANTIC REPO

OTHER EXT REPOS (GeoNames, User’s RDF stores)

EO DATA REPO

CREODIAS DATA CATALOG(OGC CSW, INSPIRE/ISO19115 METADATA)

browser.creodias.eu

ws.creodias.eu/metadata/SemanticWeb

ws.creodias.eu/metadata/findhex

sparql.creodias.eu/…

https://sparql.creodias.eu:20035/#/repositories/creodias

Hexagons

EO DATA PRODUCTS

op

tio

nal

ly

Schema.org

AdministrativeArea Island

Airport Lake

BodyOfWater Mountain

Bridge MountainRange

Building NaturalEvent

City NaturalPlace

Country ProtectedArea

Dam Region

Earthquake River

EthnicGroup Sea

Event Settlement

Glacier Volcano

CREODIAS Knowledge Database

Page 18: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

ws.creodias.eu/metadata/SemanticWeb

Page 19: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

SPARQL Endpoint

sparql.creodias.eu/…

browser.creodias.eu

Page 20: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

Semantic search - distributed query to

three RDF repositoriesState Register of Geographical Names (SRGN), CREODIAS AND DBpedia

SELECT DISTINCT ?title ?geom ?river{

{SERVICE <http://semantic.geoportal.gov.pl/sparql>

{SELECT DISTINCT ?prng_feature ?prng_label

{?prng_feature

<https://pzgik.geoportal.gov.pl/ontologies/prng/nazwaGlowna>

?prng_label.

FILTER contains(?prng_label,"Skawinka") }}}

{SERVICE <http://dbpedia.org/sparql>

{SELECT ?river

{?river a <http://dbpedia.org/ontology/River> .

?river <http://dbpedia.org/ontology/sourceCountry>

<http://dbpedia.org/resource/Poland> .

?river <http://www.w3.org/2000/01/rdf-schema#label> ?label.

FILTER contains (?label, ?prng_label)}}}

?feature a <http://ws.eodias.eu/metadata/feature> .

?feature <http://ws.eodias.eu/metadata/attribute#mission>

<http://ws.eodias.eu/metadata/mission/Sentinel-1> .

?feature <http://ws.eodias.eu/metadata/attribute#productType>

<http://ws.eodias.eu/metadata/productType/GRD> .

?feature <http://ws.eodias.eu/metadata/attribute#title> ?title .

?feature <http://ws.eodias.eu/metadata/attribute#geometry> ?geom .

?hex <http://ws.eodias.eu/metadata/attribute#feature> ?feature .

?hex <http://ws.creodias.eu/metadata/object/river> ?river .

} LIMIT 100

Find SENTINEL-1 GRD imagery, in

the area of physiogeographic

object of SRGN (Skawinka river):

Page 21: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

The use of AI

to build the real estate

knowledge base

The National Centre for Research and

Development project

POIR.01.01.01-00-1274/17

Page 22: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

Estimation and prediction of real estate

prices - modelof residential premises in

Wroclaw in 2019

Average price of 1 m² of premises

Page 23: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

Price differences depending on the type of

land use (purpose)

Page 24: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

Extraction of information form oblique

images

Page 25: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

Sentiment analysis

"In 2019, the municipal company Wrocławskie Flats, which manages real estate, plans to carryout thermomodernization and connection to the heating network of 35 tenement houses in Przedmieście Oławskie, Psie Pole and Brochów. The courtyards in Przedmieście Oławskie, Ołbinand Kleczków will also be renovated. "

"The residents of one of the new buildings at ul. Barskich at the model housing estate WuWA2. The new building is working, as can be seen on the walls inside, but the tenants became particularly worried when fragments of the facade began to fall off. Some preferred to stay on the street waiting for the fire department to secure the area. "

https://gazetawroclawska.pl/

https://www.wroclaw.pl/

SENTIMENT: POSITIVE SENTIMENT: NEGATIVE

• Determining the polarization of the text• Location extraction (exact or approximate)

Enriching information on real estate by the degree of their 'popularity', assessing the attractiveness of the area

INFERENCE

Page 26: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

Training data

Machine Learning

CNN + LSTM

Model PredictionTest data

Extraction of information from text and

pictures - zoning plans

Page 27: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

“Copernicus and Artificial Intelligence”

28 January 2020

What is now, what is coming?

1. The development of Natural Language Processing

(NLP) and convolutional networks significantly

improves the extent of information and metadata

extraction from text and pictures

2. Research on the Name Entity Recognition (NER)

make it possible to increase the scope of graph

linking

3. We can expect the acceleration in sharing data

in the form of knowledge graphs

4. Research on SPARQL queries from natural language

questions should facilitate extensive use

knowledge graphs.

Page 28: Artificial Intelligence aided Geospatial Knowledge Databases. … · 2020. 2. 3. · Artificial Intelligence aided Geospatial Knowledge Databases. Creodias and other implementations

Thank you

Adam Iwaniak

[email protected]

“Copernicus and Artificial Intelligence”

28 January 2020