Resumen:
The nasa yuwe is an official language of Colombia, it is currently in danger of extinction,
nowadays advanced strategies are being promoted from different national and indigenous
organizatons such as the information technologies to seek to support the visibility of the
language and its use through computational tools.
This document describes the development and results in the adaptation of a vector space
model for information retrieval of texts written in nasa yuwe by:
Building a closed test collection of texts written in nasa yuwe, which involved:
• Field work with nasa teachers from several nearby community shelters to the city of
Popayan.
• The establishment of 97 documents written in nasa yuwe.
• The definition of 8 queries.
• The register of expert judgment about the relevance of the documents for each query.
A prototype of information retrieval system for texts written in nasa yuwe, it was developed
taking into account:
• The adaptation of a nasa yuwe tokenizer based on the Lucene .NET standard tokenizer
(version 2.9.4)
• The definition of a stopwords removal list to apply on the documents of the nasa yuwe test
collection and queries.
• Performance evaluation of the prototype through traditional measures of the research area
as the Precision – Recall Curve.
To develop this work it was observed that although the nasa yuwe, is a language in process
of description, it was possible to adapt a tokenizer and to define a stopwords removal list
for this language, in order to get a prototype of information retrieval systems for texts written
in nasa yuwe, and through perfomance evalution of this prototype was possible to see the
adaptation of the nasa tokenizer is an important task in the recovery and this Project showed
promising results in relation with the baseline, unlike the results obtained with the stopwords
removal list, there is not substantial improvements in the performance of the prototype.