Hi Boyan,
I didn't understand that I must create JAPE rule before I do everything
else.
I'll try this now.
Thank you!
Srecko
On Thu, Jul 14, 2011 at 2:13 PM, Boyan Kukushev <boyan.kukus...@ontotext.com
> wrote:
> Hi Srecko,
>
> In order to see your Topic annotations, you must create the JAPE rule that
> Philip suggested:
>
> Phase: GazTopic
> Input: Lookup
> Options: control = appelt
> Rule: Topic
> (
> {Lookup.class == "http://proton.semanticweb.org/2006/05/protont#Topic"}
> ):topic
> -->
> :topic.Topic = {rule=GazTopic, class=:topic.Lookup.class,
> inst=:topic.Lookup.inst}
>
> and put that rule just after the gazetteer phases within the GATE pipeline.
> The easiest way to do this is using the KIM GATE interface by starting
> KIM/bin/kim(.bat) gate
>
> and modifying the pipeline.
>
> You have already added the Topic annotation type to the list of allowed
> annotation types in KIM/config/nerc.properties. After you run the pipeline
> with this new resource incuded, Topic annotations should appear in the
> default
> annotation set for each document you process.
>
> To be able to use again the pipeline, you should save it, again using the
> KIM
> GATE interface - right click on the pipeline and select 'Save application
> state'. Remember to remove (or empty) the document corpus used by the
> application. You choose whether to overwrite the default KIM pipeline
> (IE.gapp) or create a new one and point KIM to use it (setting the
> corresponding property in KIM/config/nerc.properties).
>
> Hope this helps!
>
> Regards,
> Boyan
>
> P.S. What is happening exactly:
> - the gazetteer phases use pre-defined knowledge base to find specific
> 'things' in the text you process; they produce annotations of type Lookup
> - the JAPE rule would take all Lookup annotations that have the specific
> class (in your case that is
> "http://proton.semanticweb.org/2006/05/protont#Topic";) and would create a
> new
> annotation of type Topic that is fully overlapping the current Lookup
> annotation
> - the last phase in the pipeline removes all temporary annotations - the
> Lookup annotation is also a temporary annotation, but Topic (as it is added
> to
> the allowed annotations list) will not be removed.
>
> On Thursday, July 14, 2011 14:50:51 srecko joksimovic wrote:
> > I configured nerc.properties, and now I have this:
> >
> > com.ontotext.kim.KIMConstants.IE_ANN_TYPES=Abstract, Brand, ContactInfo,
> > Date, Entity, Event, GeneralTerm, KeyLocation, KeyOrganization,
> KeyPerson,
> > KeyPhrase, Location, Money, Object, Organization, Percent, Person,
> > Position, Time, Acquirement, JobTitle, Number, Topic
> >
> > then I disabled last resource in pipeline, but I still can't see Topic.
> > Maybe I didn't understand well... should I first create Jape rule, or
> this
> > is enough to see Topic?
> >
> > Best,
> > Srecko
> >
> >
> >
> > On Thu, Jul 14, 2011 at 1:15 PM, Philip Alexiev @ Ontotext <
> >
> > philip.alex...@ontotext.com> wrote:
> > > The process is described in the customization guide you mentioned.
> > >
> > > You have added this RDF to the semantic repository. This means that
> now
> > > the gazetteer will be able to match the instances described there
> > > (searching for their labels in the texts) and will create Lookup
> > > annotations, when it finds a match. The lookup process is generally
> > > only one side of the IE process. Lookup annotations are further
> > > examined by some logic to determine their validity, or they take part
> in
> > > the recognition of more complex phrases. That is why they are not
> left
> > > after the IE has finished over the document.
> > >
> > > Tip: You can disable the last resource in the pipeline and run it again
> > > over a document to see all the annotations that are created in the
> > > process - also the temporary ones. This will show you the Lookup
> > > annotations as well. You can search for your Topic instances there.
> > >
> > > Once you have the lookups, you should tell KIM that they are important
> > > for you and you want to keep them. Add the Topic annotation type to
> the
> > > * com.ontotext.kim.KIMConstants.IE_ANN_TYPES *list. This will tell
> KIM
> > > not to clear them in the end of the IE process. Now what is left for
> > > you to do, is to create a Topic annotation over the Lookup for the
> > > topic that the Gazetteer has created. You can use a simple Jape rule
> to
> > > do that:
> > >
> > >
> > > Phase: GazTopic
> > > Input: Lookup
> > > Options: control = appelt
> > >
> > > Rule: Topic
> > > (
> > >
> > > {Lookup.class == "http://proton.semanticweb.org/2006/05/protont#Topic
> "}
> > >
> > > ):topic
> > > -->
> > >
> > > :topic.Topic = {rule = Topic, class = :topic.Lookup.class,
> > >
> > > inst=:topic.Lookup.inst}
> > >
> > >
> > > This is all that you need to include your topics in the IE process and
> to
> > > be able to see them in the graphical interface.
> > >
> > > Hope this helps
> > > philip
> > >
> > >
> > > On 14 Jul 2011, at 1:58 PM, srecko joksimovic wrote:
> > >
> > > It is little hard to explain because I didn't do customisation. I took
> > > the file where one of my colleagues did it. File contains about 1200
> > > instances and has content like this:
> > >
> > > @prefix protons: <http://proton.semanticweb.org/2006/05/protons#> .
> > > @prefix protont: <http://proton.semanticweb.org/2006/05/protont#> .
> > >
> > > <
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_07c6001f-8f5c-49e1-ae3c-92c912
> > > 1b4297
> > >
> > > a protons:Alias ;
> > > <http://www.w3.org/2000/01/rdf-schema#label>
> > >
> > > "Convex Programming@en" .
> > >
> > > <
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_63786f1d-7b3c-4872-b4e0-8227de
> > > 53c015
> > >
> > > a protons:Alias ;
> > > <http://www.w3.org/2000/01/rdf-schema#label>
> > >
> > > "Document and Text Processing@en" .
> > >
> > > <http://www.lornet.org/acm-ccs/proton#J.7.1>
> > >
> > > a protont:Topic ;
> > > protons:generatedBy
> > > <http://www.lornet.org/acm-ccs/proton#TrustedSrc>
> > >
> > > ;
> > >
> > > protons:hasMainAlias
> > >
> > > <
> > >
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_dee6eec3-b503-4d3e-a98d-ecdde5
> > > d5ee1a> .
> > >
> > > <
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_ff639082-2cc4-484e-92f7-5f0f09
> > > 898d2d
> > >
> > > a protons:Alias ;
> > > <http://www.w3.org/2000/01/rdf-schema#label>
> > >
> > > "Store and Forward Networks@en" .
> > >
> > > <
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_7f00cebc-8828-4415-83b3-2eb53a
> > > e3b0b6
> > >
> > > a protons:Alias ;
> > > <http://www.w3.org/2000/01/rdf-schema#label>
> > >
> > > "Integral Equations@en" .
> > >
> > > <
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_08a8af20-311b-4a43-b996-7d4a8c
> > > 034636
> > >
> > > a protons:Alias ;
> > > <http://www.w3.org/2000/01/rdf-schema#label>
> > >
> > > "Information Filtering@en" .
> > >
> > > <
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_bb28b422-05f4-47f8-a8f5-811610
> > > a2f77f
> > >
> > > a protons:Alias ;
> > > <http://www.w3.org/2000/01/rdf-schema#label>
> > >
> > > "Surfac eFitting@en" .
> > >
> > > <
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_8b20c6a0-1b02-4c34-bd30-740d8d
> > > 150f30
> > >
> > > a protons:Alias ;
> > > <http://www.w3.org/2000/01/rdf-schema#label>
> > >
> > > "Reliability,Availability and Serviceability@en" .
> > >
> > > <
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_b6e48ecc-f1a9-44e6-8ff9-c88176
> > > 47765a
> > >
> > > a protons:Alias ;
> > > <http://www.w3.org/2000/01/rdf-schema#label>
> > >
> > > "Aerospace@en" .
> > >
> > > <http://www.lornet.org/acm-ccs/proton#B.7.3.1>
> > >
> > > a protont:Topic ;
> > > protons:generatedBy
> > > <http://www.lornet.org/acm-ccs/proton#TrustedSrc>
> > >
> > > ;
> > >
> > > protons:hasMainAlias
> > >
> > > <
> > >
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_a4fdb728-b855-4ee6-b220-b8a93b
> > > 55f9c7> .
> > >
> > > <http://www.lornet.org/acm-ccs/proton#K.7.2>
> > >
> > > a protont:Topic ;
> > > protons:generatedBy
> > > <http://www.lornet.org/acm-ccs/proton#TrustedSrc>
> > >
> > > ;
> > >
> > > protons:hasMainAlias
> > >
> > > <
> > >
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_4694954a-ba5e-4333-9ea9-8d5b94
> > > 790c4e> .
> > >
> > > <http://www.lornet.org/acm-ccs/proton#D.3.2.10>
> > >
> > > a protont:Topic ;
> > > protons:generatedBy
> > > <http://www.lornet.org/acm-ccs/proton#TrustedSrc>
> > >
> > > ;
> > >
> > > protons:hasMainAlias
> > >
> > > <
> > >
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_d59afecf-26fc-4a5b-af92-e8c994
> > > 542b23> .
> > >
> > > <http://www.lornet.org/acm-ccs/proton#G.3.16>
> > >
> > > a protont:Topic ;
> > > protons:generatedBy
> > > <http://www.lornet.org/acm-ccs/proton#TrustedSrc>
> > >
> > > ;
> > >
> > > protons:hasMainAlias
> > >
> > > <
> > >
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_33e428d6-3157-41ca-95ee-df7973
> > > 4c5a3d> .
> > >
> > > <http://www.lornet.org/acm-ccs/proton#C.1.1.2>
> > >
> > > a protont:Topic ;
> > > protons:generatedBy
> > > <http://www.lornet.org/acm-ccs/proton#TrustedSrc>
> > >
> > > ;
> > >
> > > protons:hasMainAlias
> > >
> > > <
> > >
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_2d3fd573-5a41-403b-bdf3-22332a
> > > d9d839> .
> > >
> > > <http://www.lornet.org/acm-ccs/proton#B.5.2.1>
> > >
> > > a protont:Topic ;
> > > protons:generatedBy
> > > <http://www.lornet.org/acm-ccs/proton#TrustedSrc>
> > >
> > > ;
> > >
> > > protons:hasMainAlias
> > >
> > > <
> > >
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_171d052f-553e-4990-bd15-8416d2
> > > 8f4cf1> .
> > >
> > > <
> > >
> http://www.lornet.org/acm-ccs/proton#Alias_d680fd76-0dcf-417b-9f4e-5cbda4
> > > 616b72
> > >
> > > a protons:Alias ;
> > > <http://www.w3.org/2000/01/rdf-schema#label>
> > >
> > > "Pixel Classification@en" .
> > >
> > > I added this document to owlim.ttl and imported my instances.
> > >
> > > I tried to follow document Customizing KIM 3.pdf, but as mapping has
> > > already been done, I didn't know what else to do. Maybe I should create
> > > Jape rule, or something like that, but I think that I should see Topic
> > > with or without my instances. I'm not sure, that is only my opinion.
> > >
> > > Best,
> > > Srecko
> > >
> > > On Thu, Jul 14, 2011 at 12:48 PM, Philip Alexiev @ Ontotext <
> > >
> > > philip.alex...@ontotext.com> wrote:
> > >> Can you describe the exact actions you take to add the topics to the
> IE
> > >> logic ? The exact customizations you have made to KIM.
> > >>
> > >> Thanks,
> > >> Philip
> > >>
> > >> On 14 Jul 2011, at 1:41 PM, srecko joksimovic wrote:
> > >>
> > >> Hi Philip,
> > >> with GATE is same as with Java code. I get the same annotations. I
> tried
> > >> to edit nerc.properties and add Topic to
> > >> *com.ontotext.kim.KIMConstants.IE_ANN_TYPES * list, but nothing
> > >> changed*.
> > >> *
> > >> Do I have to change something else?
> > >>
> > >> Best,
> > >> Srecko
> > >>
> > >> On Thu, Jul 14, 2011 at 12:26 PM, Philip Alexiev @ Ontotext <
> > >>
> > >> philip.alex...@ontotext.com> wrote:
> > >>> Hi Srecko,
> > >>>
> > >>> You can run the gate interface to check exactly what annotations are
> > >>> create ant their type. You can do this by running:
> > >>> bash KIM/bin/kim gate
> > >>>
> > >>> You probably use a Jape rule to match the Lookup annotations with
> > >>> class=" http://proton.semanticweb.org/2006/05/protont#Topic"; and
> are
> > >>> creating one of the entity annotations over it (the entity
> > >>> annotations are a whitelist of annotations that remain after the
> > >>> annotation process finishes, all annotations not in this list are
> > >>> removed).
> > >>>
> > >>> So check what type of annotation you are creating.
> > >>>
> > >>> If this is not the case, please provide more details how you handle
> the
> > >>>
> > >>> topic lookups.
> > >>>
> > >>> All the best,
> > >>> Philip
> > >>>
> > >>> On 14 Jul 2011, at 1:19 PM, srecko joksimovic wrote:
> > >>> > Hello Philip,
> > >>> >
> > >>> > I included my instances in KIM. When I use web UI, I see them all,
> > >>> > and
> > >>>
> > >>> everything looks ok. But when I run code like this:
> > >>> > KIMDocument kimDoc =
> > >>>
> > >>> apiCorpora.createDocument(_string_to_annotate, true);
> > >>>
> > >>> > kimDoc = apiSemAnn.execute(kimDoc);
> > >>> >
> > >>> > KIMAnnotationSet kimASet = kimDoc.getAnnotations();
> > >>> > Set typesSet = kimASet.getAllTypes();
> > >>> > Iterator iterator = typesSet.iterator();
> > >>> >
> > >>> > // show annotations of every type separately
> > >>> > while(iterator.hasNext())
> > >>> > {
> > >>> >
> > >>> > Object key = iterator.next();
> > >>> > KIMAnnotationSet kimFilteredASet =
> > >>>
> > >>> kimASet.get(String.valueOf(key));
> > >>>
> > >>> > Iterator annIterator = kimFilteredASet.iterator();
> > >>> > System.out.println(" = Annotations of type [" +
> > >>>
> > >>> String.valueOf(key) + "] :");
> > >>>
> > >>> > while(annIterator.hasNext())
> > >>> > {
> > >>> >
> > >>> > System.out.println(" -- " + annIterator.next());
> > >>> >
> > >>> > }
> > >>> >
> > >>> > }
> > >>> > System.out.println("[ Document's Typed Annotations (end)
> > >>>
> > >>> ]");
> > >>>
> > >>> > I don't see any annotation of type Topic. I see all of them when I
> > >>> > use
> > >>>
> > >>> web UI, like I said. But when I try to annotate string from Java
> > >>> application, I don't get any Topic annotations.
> > >>>
> > >>> > Could you please help me on this one?
> > >>> >
> > >>> > Best,
> > >>> > Srecko
> > >>> > _______________________________________________
> > >>> > Kim-discussion mailing list
> > >>> > Kim-discussion@ontotext.com
> > >>> > http://ontotext.com/mailman/listinfo/kim-discussion
> > >>
> > >> _______________________________________________
> > >> Kim-discussion mailing list
> > >> Kim-discussion@ontotext.com
> > >> http://ontotext.com/mailman/listinfo/kim-discussion
>
> --
> Boyan Kukushev
> Senior Software Engineer / Java Developer
> Ontotext AD @ Sirma Group Corp.
>
_______________________________________________
Kim-discussion mailing list
Kim-discussion@ontotext.com
http://ontotext.com/mailman/listinfo/kim-discussion
