[
https://issues.apache.org/jira/browse/STANBOL-3?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13054926#comment-13054926
]
Rupert Westenthaler edited comment on STANBOL-3 at 6/25/11 3:53 PM:
--------------------------------------------------------------------
I mainly agree with Olivier as long as it is about Statements consumed by the
users (e.g. the URI used for a Person detected by the opennlp-ner engine).
However I would not recommend the same for metadata (e.g. fise:extracted-from
or the dc metadata added for Enhancements. Such configuration would be really
long and it would be really tricky (for users) to change them. In addition this
would only allow to change URIs, but not to change the structure of the
generated graph (e.g. if users want to simplify the structure of the written
enhancements). For this I suggest to use the "refactor engine" and to provide
(maybe even within the default distribution of Stanbol) different sets of rules
for different usage scenarios. It should be also possible to "install"
additional sets by using the DataFileProvider functionality already available
in Stanbol.
The entityhub is schema agnostic. However it uses some properties to store
metadata for Entities and to link the metadata of an entity with the content.
In addition it is planed to provide a set of mapping rules that can convert
properties of Entities described by the typical Linked Data Ontologies (e.g.
foaf, sioc, goodRelations ...) to the schema.org counterparts. Currently I plan
to start this by using the rather primitive mapping infrastructure built in to
the Entityhub, but I would also like to use the much richer possibilities of
the /rules package for that.
If someone could help with the rules I am happy to work on that.
was (Author: rwesten):
I mainly agree with Olivier as long as it is about Statements consumed by
the users (e.g. the URI used for a Person detected by the opennlp-ner engine).
However I would not recommend the same for metadata (e.g. fise:extracted-from
or the dc metadata added for Enhancements. Clearly this could be also done by a
configuration, but changing such a configuration would be really tricky and
sometimes one would like to change (e.g. simplify) the whole structure (graph
pattern written by the engines) something that would not be possible with a
configuration anyway.
I a user want to do this I would suggest to use the "refactor engine". We could
provide different sets of rules for different usage scenarios.
The entityhub is schema agnostic. However it uses some properties to store
metadata for Entities and to link the metadata of an entity with the content.
However I plan to provide a set of mapping rules that can convert properties of
Entities described by the typical Linked Data Ontologies (e.g. foaf, sioc,
goodRelations ...) to the schema.org counterparts. Currently I plan to start
this by using the rather primitive mapping infrastructure built in to the
Entityhub, but I would also like to use the much richer possibilities of the
/rules package for that.
If someone could help with the rules I am happy to work on that.
> Clearly defined Stanbol vocabularies
> ------------------------------------
>
> Key: STANBOL-3
> URL: https://issues.apache.org/jira/browse/STANBOL-3
> Project: Stanbol
> Issue Type: Improvement
> Reporter: Enrico Daga
>
> The Stanbol software uses some RDF vocabularies with an active role in the
> code (the code refer to these vocabularies).
> For instance, FISE uses a vocabolary for Enhancement description, KReS uses a
> vocabulary for managing ontologies and rules.
> Stanbol vocabularies should:
> - be refactored when moving to stanbol using some common namespace
> - be distributed in a well-formed RDF (which means also well-documented)
> - be collected and versioned in a common svn branch (separated from the one
> of the software code)
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
