Dileepa Jayakody created STANBOL-1161:
-----------------------------------------
Summary: Entity Disambiguation using FOAF Co-reference
Key: STANBOL-1161
URL: https://issues.apache.org/jira/browse/STANBOL-1161
Project: Stanbol
Issue Type: Task
Components: Enhancement Engines, Entityhub
Reporter: Dileepa Jayakody
Entity Disambiguation using FOAF co-reference is implemented as a new Stanbol
Enhancement Engine. The project includes 2 sub-tasks.
1. Entityhub site with substantial FOAF entities :
FOAF datasources are dispersed all over the internet, and it's essential to
aggregate a sufficient FOAF dataset in-order to perform co-reference
disambiguation techniques over data. foaf-site is the entityhub site created by
indexing btc2012 dataset including FOAF data. For more information on the
foaf-site integration in Stanbol please refer the github project at [1]. The
README.md contains the steps to integrate btc2012 dataset as an entityhub-site
in Stanbol.
2. Disambiguation Engine using FOAF co-reference techniques :
The main algorithm calculates disambiguated confidence of an entity by
analyzing the the connected-ness of the entity. This is calculated by
processing the number of URI reference matches with others from suggested
entities in the content. Co-references of URI links between entities derive
their connected-ness and the highest connected entity will have more
disambiguated confidence. The other disambiguation algorithm uses
literal-matching of selected-text with foaf:name value of the entity. If
matched, the confidence is increased. The foaf disambiguation engine project
can be found at the github project [2]. To install the engine and configure
Stanbol enhancement-chain please refer the README.md of the project.
[1] https://github.com/dileepajayakody/FOAFSite
[2] https://github.com/dileepajayakody/foaf-disambiguation
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
