EntityRanker assigns to the entities retrieved for a spot a score. The score
can be based on two different measure:
- the similarity (tf-idf) between a text window around the spot and the
wikipedia article related to the entity, if in properties
rank.by.similarity is true;
- the commonness score, i.e. the probability that the target of the spot is
this entity (
p(entity|spot)), if in properties
rank.by.commonness is true.
The entity ranker also prune candidate entities with a score lower than
the value 'entity.commonness.threshold' defined in the project.properties.
WARNING: this class could be removed or radically modified
in the future