Hi Iavor, Pablo, all
Back from vacation I started today the work on integrating the dbpedia
spotlight engines into the Apache Stanbol code base. My goal is to
have achieved this by the end of this week, but lets see how things
progress.
For this work I created an own branch
http://svn.apache.org/repos/asf/incubator/stanbol/branches/dbpedia-spotlight-engines/
that in includes just the Stanbol Enhancer (meaning
"{stanbol-trunk}/enhancer" is the root element).
I have already applied the patch provided with STANBOL-706 [2,3] and
added the new engines to the reactor POM files - so that they are
included in the default maven build. I have not yet looked at the code
nor tried to install the engines within a running Apache Stanbol
instance, but I can confirm that the build succeeds.
Comments on the proposed name changes of the used module paths,
artifactIds and package names are also welcome (see comment of
STANBOL-706 [1])
Iavor, Pablo if you are available you may want to join the #stanbol
channel on IRC. If not I will use comments on STANBOL-706 to ask
questions.
best
Rupert
[2] https://issues.apache.org/jira/browse/STANBOL-706
[3] http://svn.apache.org/viewvc?rev=1374984&view=rev (commit messages
wrongly states STANBOL-704 - sorry for that)
On Mon, Jul 30, 2012 at 7:51 PM, Iavor Jelev
<[email protected]> wrote:
> Hi Harish,
>
> thanks for your interest. Regarding details on the engine, please refer
> to the following blog post (it should answer partly your first and
> second question):
>
> http://blog.iks-project.eu/dbpedia-spotlight-integration-in-apache-stanbol-2/
>
> For a functioning demo of the described engines and chains, please refer
> to the following Stanbol instance (please note that the engines are
> rather "generously" configured, so they will return all results
> currently. Filters based on confidence metrics can of course be applied
> to improve precision):
>
> http://spotlight.dbpedia.org/stanbol/enhancer/chain
>
> Let me answer your second and third question too:
>
> 2) DBpedia Spotlight is a stand alone tool, which was designed to do a
> similar task to the current default Stanbol chain. We merely integrated
> it in Stanbol with this EnhancementEngines. It goes through the whole
> annotation life cycle (NER, DBpedia resources candidate selection,
> disambiguation, linking). So it basicly is an alternative to the already
> existing engines, as we all know - results differ according to a use
> case. So the more alternatives we have to combine (by that I mean
> engine), the better optimization possibilities there are. For further
> info on DBpedia Spotlight, please refer to the project Wiki
> (https://github.com/dbpedia-spotlight/dbpedia-spotlight/wiki)
>
> 3) No, it doesn't. As I mentioned - DBpedia Spotlight is a stand alone
> tool, and it has it's own indexes.
>
> best,
> Iavor
>
> Am 30.07.2012 19:29, schrieb harish suvarna:
>> Thanks lavor. These will be great of usage.
>> 1. How to use this engine? Should we form a new chain with [tika, langid,
>> namedentityextractionenhancementengine, your-new-dbpedia-enginee] for
>> languages having NLP components.?
>>
>> 2. How is this new dbpedia engine different from dbpediaLinking
>> (NamedEntityTaggingEngine) or dbpediaKeyword (KeywordLinkingEngine) that
>> already exists in Stanbol today?
>>
>> 3. does it use the dbpedia solr index provided in the IKS dev conf through
>> the USB?
>>
>> Thanks,
>> Harish
>>
>>
>>
>> On Mon, Jul 30, 2012 at 4:50 AM, Iavor Jelev
>> <[email protected]>wrote:
>>
>>> Hi all,
>>>
>>> as mentioned before Pablo and I developed 4 EnhancementEngines, which
>>> integrate the different aspects of DBpedia Spotlight in Apache Stanbol.
>>> We contributed them in JIRA, so if you're interested, give them a try.
>>> If there are any questions or improvement suggestions, I would
>>> appreciate your feedback. Here is the JIRA issue:
>>>
>>> https://issues.apache.org/jira/browse/STANBOL-706
>>>
>>> kind regards,
>>> Iavor
>>>
>>
>
--
| Rupert Westenthaler [email protected]
| Bodenlehenstraße 11 ++43-699-11108907
| A-5500 Bischofshofen
