Experimental Project By: Yingjie Hu Sathya Prasad Applications Prototype Lab, Esri Summer 2013 Redlands, CA.

ARCGIS ONLINE AS

LINKED DATA ?

Experimental Project By:

Yingjie HuSathya Prasad

Applications Prototype Lab, Esri

Summer 2013

Redlands, CA

UNDERSTANDING THE SEMANTIC WEB Web 3.0, Web of Data

W3C Standard

Sir Tim Berners-Lee

Currently, we are in Web 2.0

WEB 2.0 V.S. WEB 3.0

Web 2.0 Web 3.0

Key elements Documents Data

Connections Hyperlinks among documents

Links among data

Data Consumers Humans Humans and machines

Data Organization

Customized RDF

Data Publication Customized Linked Data

Data Retrieval Customized APIs SPARQL queries

UNDERSTANDING THE SEMANTIC WEB

Resource Description Framework

(RDF)

Ontologies

Data Organization Data Publishing

Linked Data principles:1. URI2. Http URI3. Self-descriptive4. External links

Data Retrieval

SPARQL query languageOriginal Data

A dataset on the Semantic Web

UNDERSTANDING THE SEMANTIC WEB

Dataset A

Dataset B Dataset C

EXISTING LINKED DATASETS

PROJECT

Publish a sample of ArcGIS Online metadata following the principles of the Semantic Web.

Explore the new capabilities enabled by the semantically structured data.

Explore the functionalities enabled by GeoSPARQL, OGC’s standard for geospatial data on the Semantic Web.

DATA IN ARCGIS ONLINE

ArcGIS Online is a GIS store A large number of items A variety of themes (tags) A rich amount of metadata for each item A lot of users from different backgrounds

Data access via REST API

REST APIData

consumerData

Request

Web app

Mobile app

Desktop app

A LINKED DATA PORTAL FOR ARCGIS ONLINE

RDF Converter

SPARQLendpoint

REST APIData

consumerData

Request

SPARQLquery

data

Web app

Mobile app

Desktop app

METHODOLOGY

1. Define ontologies (schema)

2. Mine items from ArcGIS Online (around 90,000 items)

3. Convert metadata of the items into RDF

4. Publish RDF data on a SPARQL endpoint

5. Use both SPARQL and GeoSPARQL to query

6. Build a prototype to consume the published data

Step 1, 2, 3 Step 4 Step 5

Prototype Data consumer

Step 6

ONTOLOGIES

Class level perspective

arcgis: Item

arcgis:Item-Type arcgis: typearcgis: isTypeOf

arcgis: titlexsd:stringarcgis: User

arcgis: owner

arcgis: isOwnerOf

arcgis: created

xsd:dateTime arcgis: baseMap

xsd:…

arcg

is: …

arcgis: operationalLayer

arcgis: username xsd:string

arcgis: fullName xsd:string

xsd:dateTime

arcgis: created

arcgis: Group

arcg

is: h

asIte

m

arcg

is: i

sIte

mOf

arcgis:

groupUser

arcgis: isUserO

f

ONTOLOGIES Entity level perspective

54ea730c3e3d4144b3ff45fc9512e3c5

arcgis:Web-Maparcgis: type

arcgis: titl

e

“Mountain Fire”

2013-07-18T06:23:06

arcg

is: cre

ated

arcgis: owner

arcgis: isOwnerOf

BSandigo

arc

gis

: base

Map

Topographic

arcgis:Map-Service

arcgis: type

arcgis: operationalLayer

Fire perimeter

arcgis: username “caldelmapas”

arcgis: fullName“Bosco Sandigo”

2012-06-24T06:18:12

arcgis: created

arcgis:Feature-Service

arcgis: type

arcgis:Web-App

arcgis: type

arcgis: owner

arcgis: isOw

nerOf

62487722f0104abe94859e5089c2a947

arcg

is: …

“…”

Esri Maps

arcg

is: h

asIte

m

arcg

is: i

sIte

mOf

arcgis:

groupUser

arcgis: isUserO

f

DATA CONVERSION

REST API

Ontologies (schema)

RDFConverter

RDF data

JSON

SPARQL ENDPOINT

DEMO

DEMO: BASIC SEARCH

USAGE SCENARIO: INTERNAL PORTAL

ArcGIS Linked Data portal

ArcGIS Online

Sales & Marketing

USAGE SCENARIO: EXTERNAL PORTAL & API

ArcGIS Linked Data portal

ArcGIS Online Apps

Custom Web and Mobile apps

Desktop app

Users

USAGE SCENARIO: FEDERATION

USAGE SCENARIO: FEDERATION

USAGE SCENARIO: FEDERATION

USAGE SCENARIO: SEMANTIC WEB Linking Existing Datasets on the Semantic Web

U.S. Census

LinkedGeoData

GeoWordnet

VISION

FUTURE WORK

A UI for customized SPARQL queries

Hierarchy and automatic reasoning

Language and translation

More process automation

Align our ontologies to foundational ontologies (e.g., DOLCE)

THANK YOU

http://esriurl.com/linkeddata

FEW DEMO IN SLIDES

DEMO: VALIDATION

Result from REST API

Result from SPARQL endpoint

DEMO: VALIDATION

Show <Web maps>

contains <“San Diego”>

AND <“County”>

NOT <“Agricultural”>

owned by <“onafetsid”>

DEMO: NEW QUERIES

Summary queries:

Show <# of Web maps>

Created by <ArcGIS Online User>

Between <7-1-2013> and <7-15-2013>

Contains <at least 1 Feature Service>

<at least 1 Map service>

Order by <# of Web maps Desc>

DEMO: NEW QUERIES

Link-tracking queries

DataNode 1

DataNode 2

DataNode 3

DEMO: NEW QUERIES

Link-tracking queries

Web maps

avgRating > 3

Group

Web mapsHas no ratings

Num of view >10,000

Group Owner

DEMO: KNOWLEDGE DISCOVERY Scenario 1: Basemap updating

Find out <users>

whose <web maps>

contains <World topology> as <BaseMap>

DEMO: KNOWLEDGE DISCOVERY Scenario 2: Most popular baseMap

Find out <top 10 baseMap>

which have been used <Most frequently>

rank by <frequency>

DEMO: KNOWLEDGE DISCOVERY Scenario 3: Analyze operational layers

Find out <Web maps>

which contains <Topology map> as <Basemap>

which contains <at least one “population” layer>

which contains <at least one “disaster” layer>

DEMO: KNOWLEDGE DISCOVERY

Scenario 4: Find out inactive users

Find out <users>

who <haven’t made any action (create or modify) >

in <one month>

who do not have <a profile image>

who do not have <a self-description>