uima-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From holmberg2...@comcast.net (Greg Holmberg)
Subject Re: AW: Using UIMA for structured data sources
Date Mon, 04 Aug 2008 22:47:44 GMT
Gert--


Ah, well, I don't know much about RDF, but you might want to take a look at some of the projects
IBM Research has done using UIMA, named entity extraction, and OWL:

    http://researchweb.watson.ibm.com/UIMA/SUKI/index.html

Their Semantic Search engine is also interesting:

    http://domino.research.ibm.com/comm/research_projects.nsf/pages/uima.semanticSearch.html

There are a lot of pieces you'll need to acquire to make this work: crawlers, adapters, file
format filters, an entity and relationship extractor, UIMA-to-RDF converter, etc.  There are
many choices both commercial and open source for each of these pieces.

Except that last one, which I think is a pretty hard problem.  You'll probably also have to
hire some computational linguists for the natural languages you want to support, since reliably
extracting facts from human-generated text is extremely difficult (if not impossible).  I'd
say that the system you describe is probably at or even beyond what researchers are attempting
today.  And I'm not aware of any commecial software that actually tries to reason on facts
extracted from natural language.

UIMA can help you process those CLOB and VARCHAR fields from your database, but probably isn't
a good match for processing INTEGER, DOUBLE, TIMESTAMP, etc.


Greg Holmberg


 -------------- Original message ----------------------
From: "Villemos, Gert" <gert.villemos@logica.com>
> Thanks for your answer. Indeed I need to read the UIMA documentation better.
>  
> We are building a system that will support Busines Intelligence applications 
> based on a data warehouse, as well as knowledge management features based on a 
> knowledge base. We are looking at UIMA for the loading into the knowledge base.
>  
> We have multiple data sources, some are completly structured. Others are 
> semi-structured (well defined fields, but main input is free text fields).  And 
> other again are completly unstructured (presentations, concept papers, etc).
>  
> The data warehouse we will use for report generation, trending and data mining.
>  
> On the knowledge base we would like to perform simple keyword search and indeed 
> Lucene is a candidate (Solr is a better candidate as it among others support 
> substitution) but we would also like to perform based reasoning, as well as 
> ontology based reasoning / derivation of knowledge. And we are therefore looking 
> at a knowledge base containing a RDF data graph, not just a flat index.
>  
> As far as I have been able to gather from the internet there has been some of 
> discussion on integrating Apache UIMA and Lucene, but no integration has 
> actually been made.
>  
> A better way of asking the question is therefore; for our knowledge base, what 
> do we use to create the RDF data graph? Should we:
>  
> 1. Split this into two separate tool chains, one for structured data and one for 
> unstructured data (based on UIMA)?
> 2. Use UIMA for structured as well as unstructured?
>  
> Gert.
>  
>  
> 
> ________________________________
> 
> Von: Greg Holmberg [mailto:holmberg2066@comcast.net]
> Gesendet: Mo 04.08.2008 23:39
> An: uima-user@incubator.apache.org
> Cc: Villemos, Gert
> Betreff: Re: Using UIMA for structured data sources
> 
> 
> 
> Gert--
> 
> 
> I'm not sure I understand what you're trying to build.  Your description is a 
> little vague.  Perhaps you could provide some use-cases?
> 
> I recommend that you read the UIMA docs and then ask any questions you still 
> have.
> 
> Be aware the UIMA is not a search engine.  If all you want to do is index some 
> documents, then maybe all you need is Apache Lucene.  For the structured side, 
> maybe you need a data warehouse.  Or maybe you just want to index some of the 
> CLOBs and VARCHARS into a search engine.  It's hard to tell from your 
> description.
> 
> 
> Greg Holmberg
> 
>  -------------- Original message ----------------------
> From: "Villemos, Gert" <gert.villemos@logica.com>
> > We have a number of data sources, some of which are fully structured,
> > other which are informal (unstructured). We would like to create a
> > central search facility covering structured as well as unstructured
> > data.
> > UIMA seems to fit the bill, but is focused on unstructured data.
> > Can/should I use it to also integrate structured data?
> > 
> > If yes, what are the modules which I must develop for the framework?
> > 
> > If no, what tools should I use in combination with UIMA to integrate
> > unstructured data?
> > 
> > Thanks,
> > Gert.
> >
> >
> > This e-mail and any attachment is for authorised use by the intended
> > recipient(s) only. It may contain proprietary material, confidential 
> information
> > and/or be subject to legal privilege. It should not be copied, disclosed to,
> > retained or used by, any other party. If you are not an intended recipient 
> then
> > please promptly delete this e-mail and any attachment and all copies and 
> inform
> > the sender. Thank you.
> >
> >
> 
> 
> 
> 
> 
> 
> This e-mail and any attachment is for authorised use by the intended 
> recipient(s) only. It may contain proprietary material, confidential information 
> and/or be subject to legal privilege. It should not be copied, disclosed to, 
> retained or used by, any other party. If you are not an intended recipient then 
> please promptly delete this e-mail and any attachment and all copies and inform 
> the sender. Thank you.
> 
> 


Mime
View raw message