[
https://issues.apache.org/jira/browse/STANBOL-1388?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Rafa Haro updated STANBOL-1388:
-------------------------------
Description:
Implement an Engine in Stanbol for disambiguating YAGO's entities. The
implementation should use a client of the service specified at STANBOL-1387,
i.e., using a client of the modified AIDA light framework. The workflow should
be the following:
- The engine should work within any chain including NLP analysis engines. Those
engines provide Text Annotations that have to be used as AIDA light
disambiguation service parameter, which expects a list of mentions along with
the original content as string
- Also, before using this disambuguation engine, an Entity Linking engine using
the YAGO site is needed for later mapping the candidates returned by the AIDA
light service with the entity annotations provided by the Entity Linking engine
- As response of the service, the disambiguation engine would get an ordered
(by confidence) list of candidates (entities) identified by their URI in the
YAGO dataset. Those candidates have then to be linked with actual entities in
the Stanbol YAGO site as Entity Annotations, using the confidence value
returned by AIDA-light for each candidate as the confidence value of the entity
annotation
was:
Implement an Engine in Stanbol for disambiguating YAGO's entities. The
implementation should use a client of the service specified at STANBOL-1387,
i.e., using a client of modified AIDA light framework. The workflow should be
the following:
-
> Stanbol AIDA-light Disambiguation Engine
> ----------------------------------------
>
> Key: STANBOL-1388
> URL: https://issues.apache.org/jira/browse/STANBOL-1388
> Project: Stanbol
> Issue Type: Sub-task
> Components: Enhancement Engines
> Reporter: Rafa Haro
> Fix For: 1.0.0
>
>
> Implement an Engine in Stanbol for disambiguating YAGO's entities. The
> implementation should use a client of the service specified at STANBOL-1387,
> i.e., using a client of the modified AIDA light framework. The workflow
> should be the following:
> - The engine should work within any chain including NLP analysis engines.
> Those engines provide Text Annotations that have to be used as AIDA light
> disambiguation service parameter, which expects a list of mentions along with
> the original content as string
> - Also, before using this disambuguation engine, an Entity Linking engine
> using the YAGO site is needed for later mapping the candidates returned by
> the AIDA light service with the entity annotations provided by the Entity
> Linking engine
> - As response of the service, the disambiguation engine would get an ordered
> (by confidence) list of candidates (entities) identified by their URI in the
> YAGO dataset. Those candidates have then to be linked with actual entities in
> the Stanbol YAGO site as Entity Annotations, using the confidence value
> returned by AIDA-light for each candidate as the confidence value of the
> entity annotation
--
This message was sent by Atlassian JIRA
(v6.2#6252)
