KBP 2013 Home
TAC KBP 2013 Spanish Slot Filling Track
The Spanish slot filling track extends English slot filling to the cross-lingual paradigm. Given a query and a large collection of English and Spanish documents, a system should extract slot fillers for the query and present them as they appear in the source document. While the reference KB and slots in the cross-lingual slot filling task are the same as those in the monolingual English slot filling task, the cross-lingual queries may come from either English or Spanish documents.
The cross-lingual Spanish slot filling query format and system response format are the same as for the monolingual English task, with slot fillers in the language of the source document that supports the filler. No translation of slot fillers is expected, except for normalization of dates. As in the English task, systems should not return more than one response for equivalent slot fillers; two slot fillers are considered equivalent if they refer to the same entity (in the case of named entities), have the same value (in the case of dates and quantities), or have the same English translation (in the case of strings, such as for per:title and per:charges). The correctness of system responses and the equivalence of slot fillers will be judged by bi-lingual annotators.
The cross-lingual slot filling task poses additional challenges not found in the monolingual task, including the ability to process Spanish text and to determine cross-lingual equivalence between slot fillers in English and Spanish.
Track CoordinatorJason Duncan (U.S. Department of Defense, email@example.com)
NIST is an agency of the
U.S. Department of Commerce
Last updated: Thursday, 25-Apr-2013 15:33:34 EDT
Comments to: firstname.lastname@example.org