Re: [Kim-discussion] help about kim

[Philip Alexiev]

No such functionality exists out of the box. You can implement it using a 
custom document handler and attach it to the system to be executed on various 
events.

Here is the part of KIM's public documentation describing the document handlers:
https://confluence.ontotext.com/display/KimDocs37EN/Custom+KIM+Document+Handlers

The document structure is represented by the KIMDocument java class. You can 
look it up in the API documentation which comes with KIM. The handlers also 
work with this class.

How do you plan to use the generated RDF? Can you share the big picture?

Best,
Philip

On Mar 11, 2013, at 4:36 PM, Lydia Khelifa  wrote:

> Hello Philip,
> 
> I would like  to thank you for your answer :).
> So i would like to annotate a corpora of documents with my wordlist (a manual 
> dictionary which is a list of word), and store the annotated documents in RDF 
> triples.
> So you say that your semantic annotations are stored in Owlim.
> Could i have the model or structure of this storage. I see your example and i 
> would like if you can of course get the model of storage that i will probably 
> transform into RDF one.
> My goal is to obtain an instance of annotation of my document corpora.
> Thank you very much.
> 
> Lydia K
> PHD student
> CNAM paris
> 
> 
> 2013/3/11 Philip Alexiev 
> Hello Lydia,
> 
> The simple answer to your question is - you can't. 
> 
> For the complete explanation, I will reveal a little more about KIM's 
> internals and how exactly the data is stored and where.
> 
> We can look at the process of annotating a document in KIM as a two step 
> process:
> 1. Annotate the document (Information Extraction phase)
> 2. Store the annotated document in the persistent store
> 
> The information extraction phase is performed by the GATE framework 
> (http://gate.ac.uk/) . It analyzes the text of the documents and recognizes 
> entities, that the system already knows about. Also can recognize new 
> entities based on rules or machine learning algorithms.  The output of this 
> process is a standard GATE document, with annotations over the content. The 
> GATE configuration KIM uses is customized, so that at the end, semantic 
> annotations are created. This means that the annotations represent an entity 
> from the semantic database and has features to relate it to this entity (uri 
> and class features).
> 
> The second stage is storing the document in a persistent store. KIM uses a 
> combination of a semantic store and a content store to most efficiently 
> achieve this goal.
> The document object (without the actual content) is stored in the semantic 
> store (Owlim). Further, all the features of this document and all the 
> relations to entities found in the document are also stored. This allows us 
> to see which documents mention which entities. Here is a sample document and 
> the information in the semantic store about this document:
> 
> 
>  
>  
>  .
>  
>  
>  .
>  
>  "Bumper North Sea oil profits 
> pose taxing questions for the Chancellor" .
>  
>  
> "99730440"^^ .
>  
>  "Bumper North Sea oil 
> profits pose taxing questions for the Chancellor" .
>  
>  "news 
> agency" .
>  
>  
>  .
>  
>  
>  .
>  
>  
>  .
>  
>  
>  .
>  
>  
>  .
>  
>  
>  .
>  
> 

Re: [Kim-discussion] help about kim

[Philip Alexiev]

Hello Lydia,

The simple answer to your question is - you can't. 

For the complete explanation, I will reveal a little more about KIM's internals 
and how exactly the data is stored and where.

We can look at the process of annotating a document in KIM as a two step 
process:
1. Annotate the document (Information Extraction phase)
2. Store the annotated document in the persistent store

The information extraction phase is performed by the GATE framework 
(http://gate.ac.uk/) . It analyzes the text of the documents and recognizes 
entities, that the system already knows about. Also can recognize new entities 
based on rules or machine learning algorithms.  The output of this process is a 
standard GATE document, with annotations over the content. The GATE 
configuration KIM uses is customized, so that at the end, semantic annotations 
are created. This means that the annotations represent an entity from the 
semantic database and has features to relate it to this entity (uri and class 
features).

The second stage is storing the document in a persistent store. KIM uses a 
combination of a semantic store and a content store to most efficiently achieve 
this goal.
The document object (without the actual content) is stored in the semantic 
store (Owlim). Further, all the features of this document and all the relations 
to entities found in the document are also stored. This allows us to see which 
documents mention which entities. Here is a sample document and the information 
in the semantic store about this document:


 
 
 .
 
 
 .
 
 "Bumper North Sea oil profits pose 
taxing questions for the Chancellor" .
 
 
"99730440"^^ .
 
 "Bumper North Sea oil 
profits pose taxing questions for the Chancellor" .
 
 "news agency" 
.
 
 
 .
 
 
 .
 
 
 .
 
 
 .
 
 
 .
 
 
 .
 
 
 .
 
 
 .
 
 
 .
 
 
 .
 
 
 .


The actual content of the document and also the positions of the annotations 
are not stored in the semantic store. When there is a need to visualize the 
document content and annotations, the content store is asked. In our case this 
is Lucene. 

So no RDF for the concrete annotations is kept in the RDF store. Just document 
level relations (document mentions entity) .

If you provide more information about your use case and your goals, we can be 
of more help.

All the best,
Philip Alexiev
Software Engineer, KIM team

On Mar 8, 2013, at 2:30 PM, Lydia Khelifa  wrote:

> hello,
> I'm PHD student at cnam paris and i'm using Kim to annotate semantically 
> coropora of documents.
> I'm interrestin

Re: [Kim-discussion] help about kim

[borislav popov]

Hi Lydia, i am forwarding this to the kim-discussion list. 
Cheers
Borislav

On Mar 8, 2013, at 2:30 PM, Lydia Khelifa  wrote:

> hello,
> I'm PHD student at cnam paris and i'm using Kim to annotate semantically 
> coropora of documents.
> I'm interresting in the way to export the semantic annotation in RDF but i 
> don't find the way to export it in this format.
> Could you please help me.
> 
> Lydia khelifa

___
Kim-discussion mailing list
Kim-discussion@ontotext.com
http://ontomail.semdata.org/cgi-bin/mailman/listinfo/kim-discussion