Dear Rupert,
On 10/28/2011 08:47 AM, Rupert Westenthaler wrote:
On 27.10.2011, at 16:59, Ali Anil SINACI wrote:
* The LMF semantic search component overlaps greatly with the recently by Anil
contributed "contenthub/search/engines/solr" component. Related to this it
would be great if Anil could have a look at [2] and check for similarities/differencies
and possible integration paths.
I had a look on the semantic search component of LMF. As you pointed it out,
LMF semantic search provides a convenient way to index any part of documents
with the help of RDFPath Language. I think that we can make use of this feature
in contenthub. As I described in my previous e-mail, currently, contenthub
indexes a number of semantic fields based on DBPedia relations. These are
hardcoded relations. RDFPath language can be used to indicate specific
semantic fields to be indexed along with the content itself. Let me describe
the thing in our mind in a scenario:
A user provides a domain ontology (e.g. music domain), submits to Entityhub to
be used in the enhancement process. Suppose the domain ontology includes vast
of information about artists, their albums etc... I assume that this ontology
does not include conceptual definitions (it only includes Abox definitions).
User writes an RDF Path Program (in LMF terminology) to indicate the fields to
be indexed when a content item has an enhancement related with any path in that
program. Suppose user submits a content item along with the RDF Path Program(s)
to be used to determine the fields to be indexed. Enhancement engines find an
entity (or lots of entities). Now, we execute the selected RDF Path Program(s)
and embed the results into the Solr representation of the content item.
If you have any other suggestions, please let me know so that we can discuss in
detail (in SRDC) before the meeting.
This is exactly what I was thinking about. Let me only add that such additional
Knowledge to be included within the Semantic Index might not only come from the
Entityhub, but also from other sources (like the CMS via the CMS adapter)
I you would like to help me with an Implementation of the RdfPathLanguage (e.g.
the Clerezza based Implementation, or maybe a Jena bases implementation) please
let me know. Help would be greatly welcome, because I have already a lot of
things on my TODO list before the Meeting in November (such as defining a
Proposal for the Stanbol Enhancement Structure).
We would like to get involved in the implementation of RDFPathLanguage
for Stanbol. We plan to work on this starting from next week. I think
you & LMF team already have a design in your mind. I will appreciate if
you could share your thoughts with us.
* The Semantic Search Inteface: The Contenthub currently defines it's own query API
(supports keyword based search as well as "field -> value" like constraints,
supports facets). The LMF directly exposes the RESTful API of the semantic Solr index. I
strongly prefer the approach of the LMF, because the two points already described above.
We think that we do not have to make a selection here. We can keep a simple
wrap-up on the Solr interface (contenthub's own query API) while providing the
Solr RESTful API as is. IMO a wrap-up on Solr interface would be beneficial. On
the other hand, in this interface we try to make use of an ontology to be used
in OntologyResourceSearchEngine. This might help to figure out new keywords
based on the subsumption hierarchy inside the ontology. However, I think this
may lead to performance issues and may not be useful at all. We can decide on
this later.
You forgot to mention one additional advantage for using the Solr RESTful API:
If we do that one could create the Semantic Index and than copy it over to some
other SolrServer without the need to run Stanbol directly on the production
infrastructure.
In general I would suggest to first focus the discussion on the unique features
we would like to provide with the Semantic Search component. I already included
three features I would like to have in my first Mail (Query preprocessing,
Entity Facets, Semantic Facets). As you now mention the
OntologyResourceSearchEngine is very relevant in relation to such features.
However adding such features must not necessarily mean to create an own query
language. One could also try to add such features directly to Solr by
implementing some Solr extensions.
Let me briefly comment in your suggestions about the semantic search.
But I am also the opinion that a semantic search interface should at least
provide the following three additional features:
1. Query preprocessing: e.g. substitute "Paris" in the query with
"http://dbpedia.org/resource/Paris";;
2. Entity Facets: if a keyword matches a Entity (e.g. "Paris" -> "dbpedia:Paris",
"dbpedia:Paris_Texas", "dbpedia:Paris_Hilton") than provide a Facet to the user over such possible
nnnnnnnnmatches;
As far as we understand, first and second features will be handled by
querying the Entityhub with the query keyword (Paris) i.e the first
entity obtained from the Entityhub will help us to recognize its type
and the other entities will be served as facet values of Paris facet.
3. Semantic Facets: if a user uses an instance of an ontology type (e.g. a
Place, Person, Organization) in a query, that provide facets over semantic
relations for such types (e.g. fiends for persons, products/services for
Organizations, nearby Points-Of-Interests for Places, Participants for Events,
…). To implement features like that we need components that provide query
preprocessing capabilities based on data available in the Entityhub, Ontonet …
. To me it seams that the contenthub/search/engines/ontologyresource component
provides already some functionality related to this so this might be a good
starting point.
Currently, we are trying to integrate an exploration mechanism like you
said above. It is also based on DBPedia ontology.
OntologyResourceEngine can be used for this purpose for the user
registered ontologies. Current implementation of this engine only
computes closures by exploiting the hierarchy in the ontology. RDFPath
Programs can also be an option at this point. With an RDF Path Program
user may specify the relations to be used in the exploration process.
But I think this means the user decides beforehand which fields should
be presented to him as exploration fields. I think this is open to
discussion.
best
Rupert
Regards,
Anil.
