ImageSnippets is a general purpose product for managing your images that uses linked data metadata for image description. Keywords become more meaningful, searching for images is enriched. More accurate descriptions can be gathered by experts across your global teams. Images can be published with your publishing intentions remaining clearly associated with your images.
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
ImageSnippets - Using Linked Data Metadata to Organize, Share and Publish your Images
1.
2. ImageSnippets™ is a new approach to the management of images
(and other digital resources) using semantic technology in the form of
linked data.
Semantically aware
applications can
take advantage of
both traditional and
RDFa metadata
distributed with the
images.
<span rel="rdfs:comment"><span property="rdfs:label" content="A sign alongside the abandoned tracks of the Gulf
Mobile and Ohio railroad. A jet aircraft flies overhead."></span></span></span><span rel="lio:hasSetting"><span
source="[dbpedia:Kentucky]">Kentucky</span></span></span><spanabout"><span rel="lio:hasInBackground">
<span typeof="dbpedia:Contrail">acontrail</span> </span></span>
3. Tagging with linked data offers improved tag management, querying and
innovative methods for the transport and re-use of an image with it's data.
4. ImageSnippets Users:
• Manage collections of resources (images, video,
documents) and they need to share or publish those
resources, often in multiple places simultaneously and
want to retain control of all of their data.
• Want to describe their images with much greater depth
and clarity (disambiguation and contextual tagging) and
ensure that these descriptions continue to persist with
the image no matter where it gets shared or posted.
6. keywords need
context and
disambiguation
the fit of the hood in red/green primer of car 130985, that
shows the gap between the hood and the body and shows a chalk line on
the cowl
cowl: The hood or hooded robe worn especially by a monk.
b. A draped neckline on a woman's garment.
2. A hood-shaped covering used to increase the draft of a chimney.
3. The top portion of the front part of an automobile body, supporting the windshield and dashboard.
4. The cowling on an aircraft.
7. In ImageSnippets, users can layer metadata as additional
knowledge about the images reveals itself in various
contexts.
Data can be added without having to re-write user
vocabularies, re-share or re-post the images. All metadata
added to images in ImageSnippets is dynamically available
through shared or embedded links to files.
So perhaps one person in a
team might identify superficial
data (it's a crab or jellyfish).
and then other specialists might identify
even more specific features– all on the same
images in the same system – searchable
and re-usable by all.
and later, a crustacean biologist subject matter expert
- located around the world can identify the species
8. ImageSnippets automatically provides common datasets such as: dbPedia, Yago,
Freebase.
But users can also define their own entities and datasets to describe their own
particular subject domain.
Previously engineered datasets can be loaded into ImageSnippets or the datasets can
evolve as part of the curation process.
The creation and
evolution of dataset
terms can be
orchestrated by an
administrator
exclusively
or with collaborative
input from a team of
users.
10. How it works:
Semantic technology
links data using RDF:
a subject, a property
and an object
The subject of the image
can either be the image or
a region in the image.
The property describes
how the keyword relates to
the image, such as:
"depicts" or “shows".
The object is like a
normal keyword phrase or
tag, such as: "Burt
Reynolds" or "Björn
Waldegärd.
RDF (Resource Description Framework)
is a language for describing data about resources
it’s construction uses URI’s
(universal resource identifiers (i.e. web addresses)
http://imagesnippets/thisImage.jpg
http://lio:depicts
http://dbpedia:Burt_Reynolds
11. Google (and other search engines) read and use semantic information found with resources
Rich Snippets
12. ImageSnippets has built in properties for giving context to keywords
But you can also
use properties from:
other sources
such as
or
design your own
13. ImageSnippets has an internal search function that sorts results by property,
a search for ‘New Orleans, Louisiana’, for example, might return:
Advanced users can write their own SPARQL queries against the triple stores
and named graphs using our own endpoint.
14. Ontologies link related information
– so searches can also return results without exact text based matches:
The returned results from this example found bird images even though the
text string ‘bird’ was not used anywhere in the image description