You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@stanbol.apache.org by srecko joksimovic <sr...@gmail.com> on 2012/08/21 10:16:27 UTC
Namespaces
Hi Rupert,
According to your previous answers, and [1], [2], and [3], I figured out
how to present enhancer output.
There are two namespaces that I didn't understand well:
"entityhub-query": "http://stanbol.apache.org/ontology/entityhub/query#", and
"rickModel": "http://www.iks-project.eu/ontology/rick/model/".
I am particulary interested int entityhub-query:score and
rickModel:entityRank. Could you please provide me some further info
about these namespaces?
I would like to know how you calculate "score" and "entityRank"?
Thank you,
Srecko
RE: Namespaces
Posted by Srecko Joksimovic <sr...@gmail.com>.
Hi Rupert,
I apologize because I forgot to add references:
[1] http://wiki.iks-project.eu/index.php/EnhancementStructure
[2]
http://incubator.apache.org/stanbol/docs/trunk/components/enhancer/enhanceme
ntstructure.html#fiseentityannotation
[3] http://incubator.apache.org/stanbol/docs/trunk/enhancementusage.html
From: srecko joksimovic [mailto:sreckojoksimovic@gmail.com]
Sent: Tuesday, August 21, 2012 10:16
To: Rupert Westenthaler; Rupert Westenthaler;
stanbol-dev@incubator.apache.org
Subject: Namespaces
Hi Rupert,
According to your previous answers, and [1], [2], and [3], I figured out how
to present enhancer output.
There are two namespaces that I didn't understand well:
"entityhub-query": "http://stanbol.apache.org/ontology/entityhub/query#
<http://stanbol.apache.org/ontology/entityhub/query> ", and
"rickModel": "http://www.iks-project.eu/ontology/rick/model/".
I am particulary interested int entityhub-query:score and
rickModel:entityRank. Could you please provide me some further info about
these namespaces?
I would like to know how you calculate "score" and "entityRank"?
Thank you,
Srecko
RE: Namespaces
Posted by Srecko Joksimovic <sr...@gmail.com>.
Hi,
Thank you very much. These answers are very helpful.
Best,
Srecko
-----Original Message-----
From: Rupert Westenthaler [mailto:rupert.westenthaler@gmail.com]
Sent: Tuesday, August 21, 2012 11:56
To: srecko joksimovic
Cc: Rupert Westenthaler; stanbol-dev@incubator.apache.org
Subject: Re: Namespaces
Hi,
On Tue, Aug 21, 2012 at 10:16 AM, srecko joksimovic
<sr...@gmail.com> wrote:
> Hi Rupert,
>
> According to your previous answers, and [1], [2], and [3], I figured out how
> to present enhancer output.
> There are two namespaces that I didn't understand well:
>
> "entityhub-query": "http://stanbol.apache.org/ontology/entityhub/query#",
This Namespace is used by properties used to represent query results
(e.g. the score of a result)
> and
>
> "rickModel": "http://www.iks-project.eu/ontology/rick/model/".
>
This Namespace is used by the domain model of the Entityhub. It still
uses "rick" - the old name of the Stanbol Entityhub component-
> I am particulary interested int entityhub-query:score and
> rickModel:entityRank. Could you please provide me some further info about
> these namespaces?
>
> I would like to know how you calculate "score" and "entityRank"?
>
"entityhub-query:score" is simple the score as provided by Apache
Solr. In case SPARQL is used the score value will most likely be
missing.
EnhancementEngines that suggest Entities do use the score to rank
suggestions relative to each other. However NOTE that the
fise:confidence value of the best suggestion is not influenced by the
"entityhub-query:score" but only how well the label of the Entity fits
the selected section in the text. Only of two Entities (e.g. Paris in
France and Paris in Texas) do both equally fit the selected text (e.g.
Paris) than the confidence of the highest confidence is assumed as 1
and the others are normalized relative to that.
"rickModel:entityRank" can be used to tell Stanbol the importance of
an Entity relative to others within a Dataset. For DBPedia we use the
number of incoming links from other Wikipages to calculate the
entityRank. This is than used during indexing to set the Solr
DocumentBoost. The DocumentBoost than also affects the Solr score of
query results and has therefore an influence on the
entityhub-query:score.
The entityRank value therefore depends on the indexing configuration
of the dataset. AFAIK entityRanks are currently only available for
DBpedia.
best
Rupert
> Thank you,
> Srecko
--
| Rupert Westenthaler rupert.westenthaler@gmail.com
| Bodenlehenstraße 11 ++43-699-11108907
| A-5500 Bischofshofen
RE: Namespaces
Posted by Srecko Joksimovic <sr...@gmail.com>.
Hi Rupert,
There is one more thing I would like to ask. I have attached two files: California.txt is enhancer result for a short text about California, and semantic_web.txt, short text about Semantic Web. Both of these are from Wikipedia pages, with default Stanbol configuration.
In the first case, I have @subject: URI, with @type: "dbp-ont:AdministrativeRegion", dbp-ont:Place", dbp-ont:PopulatedPlace", "http://www.opengis.net/gml/_Feature", "owl:Thing", "schema:Place" (for example). Each of these have corresponding EntityAnnotation. Each of those entity annotations has ds:relation, which points to text annotation which further describes this entity annotation. And, that was the logic I tried to implement. But, the other file, has only text annotations. Maybe, another text will produce entity and text annotation.
Could you please suggest me the most appropriate way for parsing results? Is the first example the most general one, and what should I consider when parsing results?
Best,
Srecko
-----Original Message-----
From: Rupert Westenthaler [mailto:rupert.westenthaler@gmail.com]
Sent: Tuesday, August 21, 2012 11:56
To: srecko joksimovic
Cc: Rupert Westenthaler; stanbol-dev@incubator.apache.org
Subject: Re: Namespaces
Hi,
On Tue, Aug 21, 2012 at 10:16 AM, srecko joksimovic
<sr...@gmail.com> wrote:
> Hi Rupert,
>
> According to your previous answers, and [1], [2], and [3], I figured out how
> to present enhancer output.
> There are two namespaces that I didn't understand well:
>
> "entityhub-query": "http://stanbol.apache.org/ontology/entityhub/query#",
This Namespace is used by properties used to represent query results
(e.g. the score of a result)
> and
>
> "rickModel": "http://www.iks-project.eu/ontology/rick/model/".
>
This Namespace is used by the domain model of the Entityhub. It still
uses "rick" - the old name of the Stanbol Entityhub component-
> I am particulary interested int entityhub-query:score and
> rickModel:entityRank. Could you please provide me some further info about
> these namespaces?
>
> I would like to know how you calculate "score" and "entityRank"?
>
"entityhub-query:score" is simple the score as provided by Apache
Solr. In case SPARQL is used the score value will most likely be
missing.
EnhancementEngines that suggest Entities do use the score to rank
suggestions relative to each other. However NOTE that the
fise:confidence value of the best suggestion is not influenced by the
"entityhub-query:score" but only how well the label of the Entity fits
the selected section in the text. Only of two Entities (e.g. Paris in
France and Paris in Texas) do both equally fit the selected text (e.g.
Paris) than the confidence of the highest confidence is assumed as 1
and the others are normalized relative to that.
"rickModel:entityRank" can be used to tell Stanbol the importance of
an Entity relative to others within a Dataset. For DBPedia we use the
number of incoming links from other Wikipages to calculate the
entityRank. This is than used during indexing to set the Solr
DocumentBoost. The DocumentBoost than also affects the Solr score of
query results and has therefore an influence on the
entityhub-query:score.
The entityRank value therefore depends on the indexing configuration
of the dataset. AFAIK entityRanks are currently only available for
DBpedia.
best
Rupert
> Thank you,
> Srecko
--
| Rupert Westenthaler rupert.westenthaler@gmail.com
| Bodenlehenstraße 11 ++43-699-11108907
| A-5500 Bischofshofen
Re: Namespaces
Posted by Rupert Westenthaler <ru...@gmail.com>.
Hi,
On Tue, Aug 21, 2012 at 10:16 AM, srecko joksimovic
<sr...@gmail.com> wrote:
> Hi Rupert,
>
> According to your previous answers, and [1], [2], and [3], I figured out how
> to present enhancer output.
> There are two namespaces that I didn't understand well:
>
> "entityhub-query": "http://stanbol.apache.org/ontology/entityhub/query#",
This Namespace is used by properties used to represent query results
(e.g. the score of a result)
> and
>
> "rickModel": "http://www.iks-project.eu/ontology/rick/model/".
>
This Namespace is used by the domain model of the Entityhub. It still
uses "rick" - the old name of the Stanbol Entityhub component-
> I am particulary interested int entityhub-query:score and
> rickModel:entityRank. Could you please provide me some further info about
> these namespaces?
>
> I would like to know how you calculate "score" and "entityRank"?
>
"entityhub-query:score" is simple the score as provided by Apache
Solr. In case SPARQL is used the score value will most likely be
missing.
EnhancementEngines that suggest Entities do use the score to rank
suggestions relative to each other. However NOTE that the
fise:confidence value of the best suggestion is not influenced by the
"entityhub-query:score" but only how well the label of the Entity fits
the selected section in the text. Only of two Entities (e.g. Paris in
France and Paris in Texas) do both equally fit the selected text (e.g.
Paris) than the confidence of the highest confidence is assumed as 1
and the others are normalized relative to that.
"rickModel:entityRank" can be used to tell Stanbol the importance of
an Entity relative to others within a Dataset. For DBPedia we use the
number of incoming links from other Wikipages to calculate the
entityRank. This is than used during indexing to set the Solr
DocumentBoost. The DocumentBoost than also affects the Solr score of
query results and has therefore an influence on the
entityhub-query:score.
The entityRank value therefore depends on the indexing configuration
of the dataset. AFAIK entityRanks are currently only available for
DBpedia.
best
Rupert
> Thank you,
> Srecko
--
| Rupert Westenthaler rupert.westenthaler@gmail.com
| Bodenlehenstraße 11 ++43-699-11108907
| A-5500 Bischofshofen