On Tue, Jan 15, 2013 at 11:40 AM, David Riccitelli <[email protected]> wrote:
>> 1. Simple enhancement of textual content
>
> The POST request is very similar to the Enhancer APIs [1], although the
> return data would be different. You're proposing to define an output
> different from the Enhancement Structure [2], like an XML/JSON format
> indexed by Entity, correct? e.g.
>
> {
>  language: "en",
>  entities: [{
>   "<about>": {
>     type: "{Person|Organization|Product|...}",
>     confidence: 1.0
>   }
>  },{
>    ...
>  }]
> }

To implement this one needs 2 things:

1. extract the interesting information from the Enhancement-Metadata.
Should be done in an PostProcessing Engine.
2. the serialization: For this part I would not give up on RDF, but
rater define a nice JSON-LD context [1] that produces JSON as shown in
the example above.

For this I think we should start with

* typical use cases (e.g. tag suggestion (with user interaction), auto
tagging , inline text annotation (like with annotate.js),  ...)
* specify annotations suitable for such scenarios
* implement (1) and (2) for those scenarios.


Regarding the "enhancer/task" API:

I see this only as a different RESTful service to access the Enhancer
Service. For some use cases the current API is more efficient while
for others the  enhancer/task API  has more appeal. If we provide both
options the users will decide in the end.

In anyway for EnhancementRequest specific parameters we need to
change/extend some APIs in the Enhancer. This was already discussed on
the list [2]. There was even a decision on how to do it and work will
start after the next Enhancer release (what will happen within a week
or so). After this changes there will be an EnhancementJob class that
can be created based on the request by the JAX-RS resource.


Regarding "enhancement pipeline"

> I'm saying pipeline and not enhancement chain as this goes a bit
> further, the pipeline can include selection/configuration of the DCE,
> selection/configuration of the renderer used for the enhancement graph
> etc., probably using a mini flow language to allow parts of the
> pipeline to depend on previous results (similar to the
> https://gist.github.com/2931050 idea).

Especially with all the new NLP processing related EnhancementEngines
added after STANBOL-733 this would for sure be a very welcome
extension.

best
Rupert

[1] http://json-ld.org/spec/latest/json-ld-syntax/#the-context
[2] http://markmail.org/message/ylzv4iipa5t3g5qs

--
| Rupert Westenthaler             [email protected]
| Bodenlehenstraße 11                             ++43-699-11108907
| A-5500 Bischofshofen

Reply via email to