IR Krzysztof Czuba's IR LAB page

Basic Information

• Project Title: Translingual IR
• Name: Krzysztof Czuba(email: kczuba@cs.cmu.edu)

Contents

• Abstract
• Proposal and Timelines
• System Description
• Experiments
• Results

Abstract

The project is an extension of my 11-741 project. The topic is Translingual Information Retrieval using different retrieval methods and models. In the first part of the project, the following methods will be applied to the English-Spanish UNICEF collection: DICT, DICT+Wordnet, GVSM, PRF, LSI. In the second part of the project, an automatically generated corpus will be used for methods requiring traning on bilingual corpora. The corpus has been obtained from the translation service provided on the Web by AltaVista which is based on the Systran MT system.

The original project will be extended with an experiment in which the DICT method is run using an automatically translated dictionary. The dictionary will be obtained by extracting all terms appearing in the UNICEF collection and translating them using SYSTRAN. This experiment will provide a baseline for comparing the results of the runs in which the whole corpus was automatically translated.

Also, a more sophisticated evaluation method will be used to assess how well all the methods perform. In particular, many runs on different partitions of the corpus in training, fine-tuning and test set will be done. The addition of the fine-tuning set is especially interesting since the previous results the parameters were tuned on the training set.

Proposal and Timelines

Results for GVSM and LSI on the finetuning set

Results on the test set