next up previous
Next: Results Up: Experiments Previous: Experiments

Data Description

We have used the TREC Spanish Corpora, produced by the Linguistic Data Consortium (LDC)[*], for our experiments. Particularly, one corpus of the TREC-5 collection which consists of 50 topics (queries) and 57,868 documents in Spanish language from the ``El Norte'' mexican newspaper was selected. The average size of vocabulary of each document is 191.94 terms. Each of the topics has associated its set of relevant documents. On average, the number of relevant documents per topic is 139.36. The documents, queries and relevance judgements (qrels) used in the experiments were all taken from TREC-5.



David Pinto 2007-05-08