You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@stanbol.apache.org by Rajan Shah <ra...@gmail.com> on 2015/06/02 14:53:43 UTC
Re: General Stanbol questions
Hi Rupert,
Thanks again for detailed answer.
I could setup the demo and look at the results. It's extremely powerful.
One quick question, what's the purpose of these bin files under resources
directory.
bionlp2004-DNA-en.bin
bionlp2004-RNA-en.bin
bionlp2004-cell_line-en.bin
bionlp2004-cell_type-en.bin
bionlp2004-protein-en.bin
With best regards,
Rajan
On Thu, May 28, 2015 at 3:52 AM, Rupert Westenthaler <
rupert.westenthaler@gmail.com> wrote:
> Hi Rajan,
>
> The demo never included any Java code.
>
> The module just provides configurations [1] and datafiles [2]. Those
> will be installed with the bundle using the Sling Installer and
> Stanbol DataFileProvider infrastructure when the bundle is installed.
> Note the <Install-Path> and <Data-Files> instructions configured for
> the maven-bundle-plugin in the pom.xml file.
>
> The demo also provides a shell script [3] the indexes eHealth related
> datasets and of corse the README explaining the demo
>
> best
> Rupert
>
> [1]
> http://svn.apache.org/repos/asf/stanbol/branches/release-0.12/demos/ehealth/src/main/resources/config/
> [2]
> http://svn.apache.org/repos/asf/stanbol/branches/release-0.12/demos/ehealth/src/main/resources/datafiles/
> [3]
> http://svn.apache.org/repos/asf/stanbol/branches/release-0.12/demos/ehealth/index.sh
>
> On Wed, May 27, 2015 at 1:27 PM, <ra...@gmail.com> wrote:
> > Hi Rupert,
> >
> > Thanks a lot for the detailed answers. Let me play a little bit further
> before I ask additional follow-up questions.
> >
> > As far as demo is concerned, I am interested in eHealth demo as it
> covers lots of items from my questions. At present, the Java code for it is
> missing. Is it possible to restore Java code for eHealth demo in 0.12
> branch?
> >
> > With best regards,
> > Rajan
> >
> > Sent from my iPhone
> >
> >> On May 27, 2015, at 6:42 AM, Rupert Westenthaler <
> rupert.westenthaler@gmail.com> wrote:
> >>
> >> Hi
> >>
> >>> On Wed, May 27, 2015 at 5:31 AM, Rajan Shah <ra...@gmail.com> wrote:
> >>> Hi,
> >>>
> >>> As I am trying to get my hands around stanbol, I have couple general
> design
> >>> questions.
> >>>
> >>> *1. Enhancement Chain firing and results*
> >>>
> >>> How to find out which enhancement chain detected which entities? One
> way, I
> >>> could see that by adding/removing particular chain. Is it possible to
> just
> >>> enable it via logging within current code?
> >>>
> >>> For ex.
> >>> I have a chain categorized-linking and would like to find out whether
> this
> >>> chain fired and labeled entities properly
> >>
> >> A enhancement chain has 1..* enhancement engines. The engines create
> >> the annotations not the chain. So your question should be what engine
> >> is creating an annotation. This information is provided by the
> >> dc:creator and dc:contributor metadata of the enhancement. See also
> >> the documentation at [1]
> >>
> >>>
> >>> *2. Categorize entities differently*
> >>>
> >>> Is it possible to categorize your detected entities as something else?
> >>> i.e. other than People, Organizations or Places
> >>>
> >>> What steps one need to take in current framework to achieve the same?
> >>
> >> You can use the Custom NER Model Extraction Engine [2].
> >> The models used in the documentation of this engine can be found at [3]
> >>
> >>>
> >>> *3. Domain specific modeling*
> >>>
> >>> Suppose, I have a small domain and various types of entities. I am
> >>> interested in
> >>>
> >>> a. analyzing various entities
> >>> b. linking them with other entities and find relations from
> dbpedia/freebase
> >>> c. infer interesting aspects using reasoning
> >>>
> >>> Is Stanbol the way to go or Marmotta? or Is it preferred to develop a
> >>> custom engine using Stanbol which uses internal components to perform
> all
> >>> of the above tasks?
> >>
> >> * Entity linking to your custom vocabulary in Stanbol
> >> * If you want to have your custom entities linked with
> >> dbpedia/freebase it is better to do that in the vocabulary. I think
> >> Google refine provided reconciliation to freebase. that could be
> >> definitely an option.
> >> * If you want to find additional entities contained in
> >> freebase/dbpedia configuring an other entity linking in Stanbol makes
> >> complete sense.
> >>
> >> Not sure what you mean with "infer interesting aspects using reasoning".
> >>
> >>>
> >>> *4. Enhance detected entities by annotation*
> >>>
> >>> Suppose, opennlp-ner detected an entity xyz. If I want to annotate this
> >>> entity with additional attributes/fields using different custom
> >>> vocabularies, what are the dev. steps I need to take?
> >>>
> >>
> >> If you just want to link Named Entities with a controlled vocabulary
> >> you can use the FST linking engine [4] with the Linking Mode set to
> >> NER (read the Linking Mode of the engines documentation). In short you
> >> will want to configure a "Apache Stanbol Enhancer Engine: FST Linking:
> >> Named Entities" for the vocabulary you want to link against.
> >>
> >>
> >>> *5. Previous demo project(s)*
> >>>
> >>> At the same time, any luck with restoring demo project(s) within 0.12
> >>> branch ? I believe, it demonstrates various aspects and it would be
> great
> >>> to have it restored.
> >>>
> >>
> >> I hope those are still functional in the 0.12 branch. No immediate
> >> plans to move them to 1.0.0 (mainly because of lack of time).
> >> Contributions are very welcome.
> >>
> >> Hope this helps
> >> best
> >> Rupert
> >>
> >>> Thanks in advance,
> >>> Rajan
> >>
> >>
> >> [1]
> http://stanbol.apache.org/docs/trunk/components/enhancer/enhancementstructure#fiseenhancement
> >> [2]
> https://stanbol.apache.org/docs/trunk/components/enhancer/engines/opennlpcustomner
> >> [3]
> http://svn.apache.org/repos/asf/stanbol/branches/release-0.12/demos/ehealth/src/main/resources/datafiles/
> >> [4]
> http://stanbol.apache.org/docs/trunk/components/enhancer/engines/lucenefstlinking
> >>
> >>
> >> --
> >> | Rupert Westenthaler rupert.westenthaler@gmail.com
> >> | Bodenlehenstraße 11 ++43-699-11108907
> >> | A-5500 Bischofshofen
> >> | REDLINK.CO
> ..........................................................................
> >> | http://redlink.co/
>
>
>
> --
> | Rupert Westenthaler rupert.westenthaler@gmail.com
> | Bodenlehenstraße 11 ++43-699-11108907
> | A-5500 Bischofshofen
> | REDLINK.CO
> ..........................................................................
> | http://redlink.co/
>
Re: General Stanbol questions
Posted by Rupert Westenthaler <ru...@gmail.com>.
On Tue, Jun 2, 2015 at 2:53 PM, Rajan Shah <ra...@gmail.com> wrote:
> I could setup the demo and look at the results. It's extremely powerful.
> One quick question, what's the purpose of these bin files under resources
> directory.
>
> bionlp2004-DNA-en.bin
> bionlp2004-RNA-en.bin
> bionlp2004-cell_line-en.bin
> bionlp2004-cell_type-en.bin
> bionlp2004-protein-en.bin
Those are the Open NLP models for Named Entity Extraction.
best
Rupert
>
> With best regards,
> Rajan
>
> On Thu, May 28, 2015 at 3:52 AM, Rupert Westenthaler <
> rupert.westenthaler@gmail.com> wrote:
>
>> Hi Rajan,
>>
>> The demo never included any Java code.
>>
>> The module just provides configurations [1] and datafiles [2]. Those
>> will be installed with the bundle using the Sling Installer and
>> Stanbol DataFileProvider infrastructure when the bundle is installed.
>> Note the <Install-Path> and <Data-Files> instructions configured for
>> the maven-bundle-plugin in the pom.xml file.
>>
>> The demo also provides a shell script [3] the indexes eHealth related
>> datasets and of corse the README explaining the demo
>>
>> best
>> Rupert
>>
>> [1]
>> http://svn.apache.org/repos/asf/stanbol/branches/release-0.12/demos/ehealth/src/main/resources/config/
>> [2]
>> http://svn.apache.org/repos/asf/stanbol/branches/release-0.12/demos/ehealth/src/main/resources/datafiles/
>> [3]
>> http://svn.apache.org/repos/asf/stanbol/branches/release-0.12/demos/ehealth/index.sh
>>
>> On Wed, May 27, 2015 at 1:27 PM, <ra...@gmail.com> wrote:
>> > Hi Rupert,
>> >
>> > Thanks a lot for the detailed answers. Let me play a little bit further
>> before I ask additional follow-up questions.
>> >
>> > As far as demo is concerned, I am interested in eHealth demo as it
>> covers lots of items from my questions. At present, the Java code for it is
>> missing. Is it possible to restore Java code for eHealth demo in 0.12
>> branch?
>> >
>> > With best regards,
>> > Rajan
>> >
>> > Sent from my iPhone
>> >
>> >> On May 27, 2015, at 6:42 AM, Rupert Westenthaler <
>> rupert.westenthaler@gmail.com> wrote:
>> >>
>> >> Hi
>> >>
>> >>> On Wed, May 27, 2015 at 5:31 AM, Rajan Shah <ra...@gmail.com> wrote:
>> >>> Hi,
>> >>>
>> >>> As I am trying to get my hands around stanbol, I have couple general
>> design
>> >>> questions.
>> >>>
>> >>> *1. Enhancement Chain firing and results*
>> >>>
>> >>> How to find out which enhancement chain detected which entities? One
>> way, I
>> >>> could see that by adding/removing particular chain. Is it possible to
>> just
>> >>> enable it via logging within current code?
>> >>>
>> >>> For ex.
>> >>> I have a chain categorized-linking and would like to find out whether
>> this
>> >>> chain fired and labeled entities properly
>> >>
>> >> A enhancement chain has 1..* enhancement engines. The engines create
>> >> the annotations not the chain. So your question should be what engine
>> >> is creating an annotation. This information is provided by the
>> >> dc:creator and dc:contributor metadata of the enhancement. See also
>> >> the documentation at [1]
>> >>
>> >>>
>> >>> *2. Categorize entities differently*
>> >>>
>> >>> Is it possible to categorize your detected entities as something else?
>> >>> i.e. other than People, Organizations or Places
>> >>>
>> >>> What steps one need to take in current framework to achieve the same?
>> >>
>> >> You can use the Custom NER Model Extraction Engine [2].
>> >> The models used in the documentation of this engine can be found at [3]
>> >>
>> >>>
>> >>> *3. Domain specific modeling*
>> >>>
>> >>> Suppose, I have a small domain and various types of entities. I am
>> >>> interested in
>> >>>
>> >>> a. analyzing various entities
>> >>> b. linking them with other entities and find relations from
>> dbpedia/freebase
>> >>> c. infer interesting aspects using reasoning
>> >>>
>> >>> Is Stanbol the way to go or Marmotta? or Is it preferred to develop a
>> >>> custom engine using Stanbol which uses internal components to perform
>> all
>> >>> of the above tasks?
>> >>
>> >> * Entity linking to your custom vocabulary in Stanbol
>> >> * If you want to have your custom entities linked with
>> >> dbpedia/freebase it is better to do that in the vocabulary. I think
>> >> Google refine provided reconciliation to freebase. that could be
>> >> definitely an option.
>> >> * If you want to find additional entities contained in
>> >> freebase/dbpedia configuring an other entity linking in Stanbol makes
>> >> complete sense.
>> >>
>> >> Not sure what you mean with "infer interesting aspects using reasoning".
>> >>
>> >>>
>> >>> *4. Enhance detected entities by annotation*
>> >>>
>> >>> Suppose, opennlp-ner detected an entity xyz. If I want to annotate this
>> >>> entity with additional attributes/fields using different custom
>> >>> vocabularies, what are the dev. steps I need to take?
>> >>>
>> >>
>> >> If you just want to link Named Entities with a controlled vocabulary
>> >> you can use the FST linking engine [4] with the Linking Mode set to
>> >> NER (read the Linking Mode of the engines documentation). In short you
>> >> will want to configure a "Apache Stanbol Enhancer Engine: FST Linking:
>> >> Named Entities" for the vocabulary you want to link against.
>> >>
>> >>
>> >>> *5. Previous demo project(s)*
>> >>>
>> >>> At the same time, any luck with restoring demo project(s) within 0.12
>> >>> branch ? I believe, it demonstrates various aspects and it would be
>> great
>> >>> to have it restored.
>> >>>
>> >>
>> >> I hope those are still functional in the 0.12 branch. No immediate
>> >> plans to move them to 1.0.0 (mainly because of lack of time).
>> >> Contributions are very welcome.
>> >>
>> >> Hope this helps
>> >> best
>> >> Rupert
>> >>
>> >>> Thanks in advance,
>> >>> Rajan
>> >>
>> >>
>> >> [1]
>> http://stanbol.apache.org/docs/trunk/components/enhancer/enhancementstructure#fiseenhancement
>> >> [2]
>> https://stanbol.apache.org/docs/trunk/components/enhancer/engines/opennlpcustomner
>> >> [3]
>> http://svn.apache.org/repos/asf/stanbol/branches/release-0.12/demos/ehealth/src/main/resources/datafiles/
>> >> [4]
>> http://stanbol.apache.org/docs/trunk/components/enhancer/engines/lucenefstlinking
>> >>
>> >>
>> >> --
>> >> | Rupert Westenthaler rupert.westenthaler@gmail.com
>> >> | Bodenlehenstraße 11 ++43-699-11108907
>> >> | A-5500 Bischofshofen
>> >> | REDLINK.CO
>> ..........................................................................
>> >> | http://redlink.co/
>>
>>
>>
>> --
>> | Rupert Westenthaler rupert.westenthaler@gmail.com
>> | Bodenlehenstraße 11 ++43-699-11108907
>> | A-5500 Bischofshofen
>> | REDLINK.CO
>> ..........................................................................
>> | http://redlink.co/
>>
--
| Rupert Westenthaler rupert.westenthaler@gmail.com
| Bodenlehenstraße 11 ++43-699-11108907
| A-5500 Bischofshofen
| REDLINK.CO ..........................................................................
| http://redlink.co/