Semantic Web Challenges for Visualisation and Visual Analytics

11
Konstanz April 20111 Semantic Web Challenges for Visualisation & VA Alan Dix Lancaster University and Talis hcibook.com/alan alandix.com/blog

description

Talk at retreat of Daniel Kiem's Konstanz visualisation research group. The tal

Transcript of Semantic Web Challenges for Visualisation and Visual Analytics

Page 1: Semantic Web Challenges for Visualisation and Visual Analytics

Kons

tanz

April

201

11

Semantic Web Challenges for Visualisation & VA

Alan DixLancaster University

and Talis

hcibook.com/alanalandix.com/blog

Page 2: Semantic Web Challenges for Visualisation and Visual Analytics

Kons

tanz

April

201

11

Semantic Web – What is it?

• web of data for computation

• technologies: RDF, OWL, triples and ontologies– everything comes in threes

<http://alandix.com/me> <foaf:name> “Alan Dix” .<http://alandix.com/me> <bz:works_at>

<http://talis.com/> .

• linking (open) data

Page 3: Semantic Web Challenges for Visualisation and Visual Analytics

Kons

tanz

April

201

11

LOD cloud

Page 4: Semantic Web Challenges for Visualisation and Visual Analytics

Kons

tanz

April

201

11

three paths to Sem Web

SemWeb:RDF,etc.

<fresh><semantic>

<data>

from HTMLadd markup (RDFa)

or data detectors

from existing data(CSV, RDMS, etc.)

Page 5: Semantic Web Challenges for Visualisation and Visual Analytics

Kons

tanz

April

201

11

from raw data to semantic data

<RDF><triples>

existingraw data

linkedopen data

convert /describe

linkage(via URIs)

Page 6: Semantic Web Challenges for Visualisation and Visual Analytics

Kons

tanz

April

201

11

existing raw data

• understanding:data, domain, connections

• focus on structure• heterogeneous representations• different views (for different purposes)

de-normalised• sub-unit semantics (e.g. “1000 kp rising” )

• super-unit semantics (e.g. lat&long)

<RDF>

Page 7: Semantic Web Challenges for Visualisation and Visual Analytics

Kons

tanz

April

201

11

converting / describing

• identity:are two things in different places the same

• rules and exceptionslarge so try to do it with rules (e.g. natural

keys)but need exceptions when rules don’t hold

... finding them – outliers

• combination of hand-craft and crafted automation= visual analytics !!

<RDF>

Page 8: Semantic Web Challenges for Visualisation and Visual Analytics

Kons

tanz

April

201

11

RDF / Semantic Data

• schema-lessvocabulary, but optional schema

• potentially rich class & predicate typescoloured graph, not just tables!sometimes level mixing (meta-instance)and maybe BIG

• sometimes text + structure (e.g. ODP)but many small text units (unlike classic IR)

<RDF>

Page 9: Semantic Web Challenges for Visualisation and Visual Analytics

Kons

tanz

April

201

11

linkage

• establishing identity• value relationships

• similar issues to raw=>RDF transformation

<RDF>

Page 10: Semantic Web Challenges for Visualisation and Visual Analytics

Kons

tanz

April

201

11

linked open data

• heterogeneous• incomplete• pluggable visualisations• URI links ... but also value relationships• very BIG – the web of data

<RDF>

Page 11: Semantic Web Challenges for Visualisation and Visual Analytics

Kons

tanz

April

201

11

let me know what you do!