[
https://issues.apache.org/jira/browse/STANBOL-107?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13000987#comment-13000987
]
Enrico Daga commented on STANBOL-107:
-------------------------------------
Some examples should be setup from existing engines, as ogrisel suggester.
Don't do it now since there are lot of ways to do it and I - personally - want
to think more about it.
> semantic description of the engines
> -----------------------------------
>
> Key: STANBOL-107
> URL: https://issues.apache.org/jira/browse/STANBOL-107
> Project: Stanbol
> Issue Type: New Feature
> Components: Enhancer
> Reporter: Enrico Daga
> Priority: Minor
>
> It would be nice to find a way to let engines declare which is the
> contribution they are going to provide.
> I see at least the following kinds of enhancements:
> 1) tagging: detect keywords, entities, concepts "within" the content
> 2) categorization/classification: locate the content in a conceptual place
> within a given framework. For example an engine could state that the document
> has "Secon World War" as primary topic, or "Theatre" in the framework of
> DBPedia categories, or state that is an E-mail, or a News, in the framework
> of the CMS document types;
> 3) metadata: the engine extracts metadata from within the content. For
> instance it returns the PDF metadata in RDF using the dublin core vocabulary
> 4) embedded knowledge: the source document is a rich HTML (with RDFa,
> Microformats) or it is a structured file (why not an RDF file, say a FOAF
> profile?)
> then, the enhancement engine should also say HOW it contributes in terms of
> vocabulary elements
> 1) Does the engine add annotation roles?
> 2) Does it add entity types?
> 3) Which metadata fields it will return?
> This could be done with an RDF description stating which are the terms the
> engine will introduce in relation to the ones of the Stanbol Enhancement base
> ontology (STANBOL-52). This is also related to STANBOL-3.
--
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
