Rbosaghz writesup Wang Cohen 2007

This is a review of Wang_2007_language_independent_set_expansion_of_named_entities_using_the_web by user:Rbosaghz

This paper is about Set Expansion, which is directly tied to bootstrapping, a topic we've been looking at for the past few weeks. The task is to take a set of named entities and expand the set into more named entities of the same "type".

This paper decribes the SEAL system, which tries to tackle set expansion problem. They compare to Google Sets, and perform better in addition to being able to hand multiple languages.

Instead of a review, I ask the following:

The set expansion problem in its current version is ill-defined. Growing a list of named-entities to include other named-entities of the same "type" is inherently ambiguous, since all named entities are at least named entities, and thus have something in common. On the other extreme end, all named entities are the only entities of their own type (except for real synonyms). So it seems the set-expansion problem is a thinly supervised clustering function (on the set of all named entities), without a well-defined objective function, to help disambiguate different clusterings.