Hi Rupert,
see my comments inline
Am 25.07.2012 10:00, schrieb Rupert Westenthaler:
Hi Sebastian,
## Points of Interest (for Stanbol Developers/Users)
1. NIF URI scheme: Basically encoding information of
fise:TextAnnotation within the URI (see [1] for details). Two variants
*
"{content-item-uri}#offset_{startindex}_{endindex}_{max-20chars-of-selected-text}"
*
"{content-item-uri}#hash_{context-length}_{length-selected}_{context-md5}_{max-20chars-of-selected-text}"
This could be relatively easy implemented by adding support for
those URIs to the EnhancementEngineHelper#createTextAnnotation(..)
methods. Note that this would require an API change because
EnhancementEngines would need to parse selected-text, context and
offset values to correctly calculate the NIF compliant URI
I can only assume that FISE and NIF play along well. I would need to see
some turtle or owl axioms, not just fancy images.
2. Ontologies of Linguistic Annotations (OLiA) provides URIs for
things like "olia:Noun", "olia:Verb". It might also be useful to
express Lemmas and other features (e.g. as provided by the CELI
engines). However this ontology is also quite big and therefore hard
to grasp.
If this could be used to determine if a POS (Part of Speech) tag
provided by some NLP tool for some language corresponds to a "Noun",
"Verb" ... it could be really useful. Currently the
KeywordLinkingEngine manages those information as part of its
configuration. But as I mentioned above - I do not unstained OLiA good
enough the be sure if such a thing is possible/feasible.
This is exactly what OLiA is for. Via reasoning you can conlude that
these three tags from different sets
http://purl.org/olia/stts.owl#ADJA
http://purl.org/olia/brown.owl#JJ#
http://purl.org/olia/susa.owl#JBy
are all of type
http://purl.org/olia/olia.owl#Adjective
Note that we are currently migrating servers.
So you can look at the the ontologies here http://nlp2rdf.lod2.eu/olia/
3. NIF and Open Annotation Core Data Model (OA) [2] (related to
STANBOL-351): There was recently the suggestion to adopt OA for Apache
Stanbol and NIF seams to have quite an overlap with OA. Some more
information about that would clearly help.
Yes, there is some overlap. OpenAnnotation is much more elaborate in a
good and a bad way. So for one NIF triple they add a lot of overhead:
http://www.w3.org/community/openannotation/wiki/TextCommentOnWebPage
I really wonder how they are planning to generate all these UUID's and
whether this approach is feasible for NLP.
I am not sure, whether it will get adopted so well. NIF 2.0 will be
compatible with http://www.w3.org/TR/prov-aq/ and there will be a
transformation also to OpenAnnotation
Finally I would really like if some one could actually translate the
FISE annotations depicted in [3] to NIF. I think this would make it
much easier for members of the Stanbol Community to grasp NIF.
Including information like
I really would like to see some turtle or OWL from your side. I could
not find *anything* . I feel unable to translate an image such as [3] to
NIF. We should really have a telco, as we are having a deadlock. I wrote
my initial email in the first place, because I could not find proper
documentation on your pages. I see that somebody answered on the google
doc, where i put some questions:
https://docs.google.com/document/d/15lNMJ3owfKmJX-DuHDJVn4t3aRgYcw-DxKZs2-aXtwU/edit
I asked for an example of FISE there and the answer was: Every result of
an enhancement request , which is not helpflul. I would really need to
see some more details and ideally talk to somebody. My Skype account is
"sebastian.hellmann"
I am updating the NLP2RDF wiki during the next weeks.
All the best,
Sebastian
* POS tags for "Bob" and "Marley"
* Chunk "Bob Marley"? Can Chunks be connected to Words?
* Are EntityAnnotations within the scope of NIF? If yes, how would
they encoded by using NIF
* How to express metadata (e.g. dc:creator, fise:extracted-from,
fise:confidence) in NIF
best
Rupert
[1] http://svn.aksw.org/papers/2012/WWW_NIF/public/string_ontology.pdf
[2] http://www.openannotation.org/spec/core/
[3]
http://incubator.apache.org/stanbol/docs/trunk/components/enhancer/enhancementstructure.html#overview-on-the-stanbol-enhancement-structure
On Wed, Jul 4, 2012 at 9:06 AM, Sebastian Hellmann
<[email protected]> wrote:
Hi Rupert,
I found the dead links here:
http://incubator.apache.org/stanbol/docs/trunk/enhancementusage.html
Basically every link with "fise:" at the beginning, see the highlighed part
here:
http://pcai042.informatik.uni-leipzig.de/~swp12-9/vorprojekt/index.php?annotation_request=http%3A//incubator.apache.org/stanbol/docs/trunk/enhancementusage.html%23frag_f0935e4cd5920aa6c7c996a5ee53a70f
I will have a look at the ontology soon.
All the best,
Sebastian
Am 04.07.2012 08:22, schrieb Rupert Westenthaler:
Hi Sebastian, all
Thanks for this mail/proposal. It is definitely well received by
myself and I think also the Stanbol Community as a whole.
This is only a quick replay to the question about the wrong URL for
the Enhancement Structure Documentation. For a detailed replay I will
definitely need more time.
On Wed, Jul 4, 2012 at 1:33 AM, <[email protected]> wrote:
I got a 404 on
http://incubator.apache.org/enhancer/enhancementstructure.html
I read "fise" somewhere. What is it? How does it compare to NIF? What URIs
do you use? How many triples do you have per annotation?
This looks like a link to that page started unintentional with a '/'.
Can you remember the occurrence of this link?
The correct URL is
http://incubator.apache.org/stanbol/docs/trunk/enhancer/enhancementstructure.html
Regarding typical use cases you should also have a look as this usage
scenario
http://incubator.apache.org/stanbol/docs/trunk/enhancementusage.html
The Ontologies can be found on the SVN (we will make them
de-referenceable as soon as we are a full Apache Project and do own
the URLs)
http://svn.apache.org/repos/asf/incubator/stanbol/trunk/enhancer/generic/servicesapi/src/main/resources/
best
Rupert
--
Dipl. Inf. Sebastian Hellmann
Department of Computer Science, University of Leipzig
Events: http://wole2012.eurecom.fr (*Deadline: July 31st 2012*)
Projects: http://nlp2rdf.org , http://dbpedia.org
Homepage: http://bis.informatik.uni-leipzig.de/SebastianHellmann
Research Group: http://aksw.org
--
| Rupert Westenthaler [email protected]
| Bodenlehenstraße 11 ++43-699-11108907
| A-5500 Bischofshofen
--
Dipl. Inf. Sebastian Hellmann
Department of Computer Science, University of Leipzig
Events:
* http://sabre2012.infai.org/mlode (Leipzig, Sept. 23-24-25, 2012)
* http://wole2012.eurecom.fr (*Deadline: July 31st 2012*)
Projects: http://nlp2rdf.org , http://dbpedia.org
Homepage: http://bis.informatik.uni-leipzig.de/SebastianHellmann
Research Group: http://aksw.org
