27
Civet From content to Linked Open Data Sebastian Ryszard Kruk, PhD Arkadiusz Kwoska Friday, December 14, 12

Civet - from Content to Linked Open Data

Embed Size (px)

DESCRIPTION

A presentation about Civet, a service delivered by Knowledge Hives, we (Arek and Sebastian) gave on June 8th at SemTech 2011. The presentation also mentioned results of the "Semantic tools for digital libraries" project a.k.a. SemLib, which is a 24th month R&D project supported by EU FP7 Theme: Research for SMEs (no. FP7-SME-2010-01-262301-SEMLIB) commenced in January 2011. More info at http://www.semlibproject.eu

Citation preview

Page 1: Civet - from Content to Linked Open Data

CivetFrom content to Linked Open Data

Sebastian Ryszard Kruk, PhDArkadiusz Kwoska

Friday, December 14, 12

Page 2: Civet - from Content to Linked Open Data

CIVET - from text to RDFa

Current issues

where to get

vocabularies for machines

how to make text

better understood by

machines

OpenVocabulary - publishing dictionaries

Potential problems

Where to go from

here

Friday, December 14, 12

Page 3: Civet - from Content to Linked Open Data

Human text vs

machines

plain text is virtually

useless for machines

source: http://www.flickr.com/photos/libaer2002/2398312710/

Friday, December 14, 12

Page 4: Civet - from Content to Linked Open Data

Human text vs

machines continued

but not too many

orders

NLP and tags only lower the

order of magnitude of

the problem

source: http://www.flickr.com/photos/matthias_haas67/4200170217/

Friday, December 14, 12

Page 5: Civet - from Content to Linked Open Data

Human text vs

machines continued

but there’s a lot

of human knowledge in just plain

text

we are already pass retrieving the low “hanging fruit”

semantics from databases

source: http://www.flickr.com/photos/katrinlorenzen/5409008140

Friday, December 14, 12

Page 6: Civet - from Content to Linked Open Data

CIVET - from text to RDFa

Current issues

where to get

vocabularies for machines

how to make text

better understood by

machines

OpenVocabulary - publishing dictionaries

Potential problems

Where to go from

here

Friday, December 14, 12

Page 7: Civet - from Content to Linked Open Data

Where to get LOD Vocabs ?

source:http://www.flickr.com/photos/bfurlong/2351689062/

Friday, December 14, 12

Page 8: Civet - from Content to Linked Open Data

Where to get LOD

Vocabs ? continued

Friday, December 14, 12

Page 9: Civet - from Content to Linked Open Data

Where to get LOD

Vocabs ? continued

Friday, December 14, 12

Page 10: Civet - from Content to Linked Open Data

CIVET from text to

RDFa

Current issues

where to get

vocabularies for machines

how to make text

better understood by

machines

OpenVocabulary - publishing dictionaries

Potential problems

Where to go from

here

Friday, December 14, 12

Page 11: Civet - from Content to Linked Open Data

CIVET - from text to RDFa

source:http://www.flickr.com/photos/xjrlokix/3269530621/

Friday, December 14, 12

Page 12: Civet - from Content to Linked Open Data

CIVET - from text to RDFa

continued

source: http://www.flickr.com/photos/72213316@N00/3149423057/

Friday, December 14, 12

Page 13: Civet - from Content to Linked Open Data

CIVET - from text to RDFa

continued

Friday, December 14, 12

Page 14: Civet - from Content to Linked Open Data

Friday, December 14, 12

Page 15: Civet - from Content to Linked Open Data

Friday, December 14, 12

Page 16: Civet - from Content to Linked Open Data

Friday, December 14, 12

Page 17: Civet - from Content to Linked Open Data

Friday, December 14, 12

Page 18: Civet - from Content to Linked Open Data

CIVET - from text to RDFa

Current issues

where to get

vocabularies for machines

how to make text

better understood by

machines

OpenVocabulary - publishing dictionaries

Potential problems

Where to go from

here

Friday, December 14, 12

Page 19: Civet - from Content to Linked Open Data

OpenVocabulary: from legacy

dictionaries to LOD

Friday, December 14, 12

Page 20: Civet - from Content to Linked Open Data

CIVET - from text to RDFa

Current issues

where to get

vocabularies for machines

how to make text

better understood by

machines

OpenVocabulary - publishing dictionaries

Potential problems

Where to go from

here

Friday, December 14, 12

Page 21: Civet - from Content to Linked Open Data

Mind the hole ...

Mind the hole

Friday, December 14, 12

Page 22: Civet - from Content to Linked Open Data

Mind the hole

continued

working with text is

sometimes tricky

often, when you

deal with non-English text

Friday, December 14, 12

Page 23: Civet - from Content to Linked Open Data

Mind the hole

continued

current meaning

discovery algorithms are still

questionable

but they do not take into account linked data  

Friday, December 14, 12

Page 24: Civet - from Content to Linked Open Data

CIVET - from text to RDFa

Current issues

where to get

vocabularies for machines

how to make text

better understood by

machines

OpenVocabulary - publishing dictionaries

Potential problems

Where to go from

here

Friday, December 14, 12

Page 25: Civet - from Content to Linked Open Data

Where to go from

here?

Improve, improve, improve ... meaning discovery

quality and performance

discovering key

meaningful facts

Friday, December 14, 12

Page 26: Civet - from Content to Linked Open Data

Where to go from

here?continued

Validate, validate, validate ... results

with real users with

real problems

SemLib annotation engine

for gathering feedback discovering

key meaningful facts

SemLib

recommendation engine to utilize results

Friday, December 14, 12

Page 27: Civet - from Content to Linked Open Data

CivetFrom content to Linked Open Data

http://civet.knowledgehives.com/Semantic tools for digital libraries a.k.a. SemLib is a 24th month R&D project supported by EU FP7 Theme: Research for SMEs (no. FP7-SME-2010-01-262301-SEMLIB) commenced in January 2011.

More info at http://www.semlibproject.eu

Friday, December 14, 12