stanbol-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rupert Westenthaler <rupert.westentha...@gmail.com>
Subject Re: Etcware early adoption proposal
Date Fri, 02 Mar 2012 11:35:49 GMT
Hi Andrea


On 29.02.2012, at 16:08, Andrea Ciapetti wrote:
> 
> Our company has also acquired significant competences in the usage of
> semantic technologies and standards. After the experience in an Italian
> research project, we have developed a product for SKOS thesaurus publishing
> and management, named SKOSware (http://www.skosware.it).
> 
was not able to look this up, because the website seems do be down at the moment.
> 
> Our vision is to integrate Stanbol in place of our manual metadata
> enrichment for the CMS contents. This will allow us to add additional
> content enrichment through Stanbol engines. Moreover, content enrichment
> and tagging will become mostly automatic in this way. Stanbol integration
> in our Liferay solution will be “loosely coupled” to allow an easy porting
> in the next version of the CMS, and to enable a maximum degree of reuse of
> our semantic customization.

nice use case. Do you also plan to use extracted enhancements as suggestions to extend the
thesaurus managed by your tool? I am personally interested in the feasibility of such use
cases.

> *The solution will be integrated in the Italian data protection Authority
> portal as a demo*, running Stanbol enhancement engines on their document
> corpus composed by 12.000 items, 2.000 of which already manually enriched
> with metadata.
> 
> Our plan to integrate Stanbol is based on the following steps:
> 
>   1. Thesauri selected from SKOSware are imported into Stanbol to create a
>   base custom knowledge domain.

Currently this is only possible by using the Entityhub Indexing tool. This is OK for one-time
imports and sporadic updates, but it might not be sufficient for your use case. So if you
have additional requirements we might need to add some new functionality.

You might also want to have a look at EnhancementChains. This would allow you to configure
multiple enhancement endpoints that use a different set of Thesauri. The documentation of
the Stanbol Enhancer [1] provides more information on that

[1] http://incubator.apache.org/stanbol/docs/trunk/enhancer/

>   2. The Content editor creates or updates contents and documents on
>   Liferay. These contents are enriched through Stanbol enhancement engines,
>   on editing post-process event.
>   3. The Liferay administrator launches Stanbol automatic metadata
>   enrichment for all contents and documents (batch enrichment process).
>   4. The End-user searches contents and documents by using full-text
>   search or tag-cloud-based search and refines the results or expands the
>   search scope on similar or related contents (under the scenes SKOS
>   thesaurus concepts and semantic relations are used to define the related
>   contents).

The Stanbol Contenthub allows to build Semantic Indexes based on Enhancement Results. This
components basically allows you to configure you own SemanticIndex layout (by using LDPath[2]).
Queries to the managed semantic index directly use Apache Solr [3]. 
Note that we do plan to provide considerable improvements to this components in the coming
months [4]. 

[2] http://code.google.com/p/ldpath/
[3]  http://lucene.apache.org/solr/
[4] https://issues.apache.org/jira/browse/STANBOL-471

>   5. As the end-user views portal contents, terms similar to SKOS concepts
>   (skos:prefLabel or skos:altLabel are used for entity highlighting) are
>   automatically decorated and their description is shown on some specific GUI
>   event (like mouseover).

You might be interested in nano "annotate.js" [5]. Try the demo at [6]

[5] https://github.com/szabyg/annotate.js
[6] http://dev.iks-project.eu:8081/enhancervie

>   6. Inference rules and semantic reasoning will be used to complete and
>   enrich the domain knowledge base, thus suggesting additional concepts and
>   OWL relations.

Stanbol includes support for rules and reasoning. However I am not an expert with that.

Simple reasoning things can be also implemented by using LDPath directly on the Entityhub.
You can try 

    http://dev.iks-project.eu:8081/entityhub/site/gemet/find

search e.g. for 

    nuclear

and use the following LDPath

skos:prefLabel;
skos:altLabel;
skos:hiddenLabel;
rdfs:label = (skos:prefLabel | skos:altLabel | skos:hiddenLabel);
skos:notation

skos:inScheme;

skos:broader = (skos:broader | ^skos:narrower);
skos:broaderTransitive = (skos:broader | ^skos:narrower)+;

skos:narrower = (^skos:broader | skos:narrower);
skos:narrowerTransitive = (^skos:broader | skos:narrower)+;

skos:related = (skos:related | skos:relatedMatch);
skos:relatedMatch;
skos:exactMatch = (skos:exactMatch)+;
skos:closeMatch = (skos:closeMatch | (skos:exactMatch)+);
skos:broaderMatch = (^skos:narrowMatch | skos:broaderMatch);
skos:narrowMatch = (skos:narrowMatch | ^skos:broaderMatch);


This will provide you with SKOS concepts that match the search term but also include information
such as the transitive closure for broaderTransitive and narrowerTransitive

>   7. Optional use of some IKS VIE widgets on the frontend presentation
>   layer.

Ok you are already aware of VIE ^^

best
Rupert


Mime
View raw message