
A Machine Translation Approach to Cross Language Text Retrieval
Cross Language Text Retrieval (CLTR) has been defined as the retrieval of documents in a language different from that of the original query. To make this possible some kind of mechanism has to be applied in order to translate the information contained in the source sentence. Many different approaches have been carried out with the purpose of transferring the information from the source language query to the target language one. Though all these methods deal with a way of translating as much information as possible from the source query, little research has been conducted in relation to the field of Machine Translation (MT). The purpose of this research work is to determine the feasibility of using MT techniques for CLTR. Specifically, I will describe how a MT system has been adapted without much effort to translate Spanish queries of a specific domain, i.e. Finance and Economics, into English in order to retrieve documents related to that field. The results of this process will then be compared with the results obtained from the retrieval of the original English queries. Thus, I will discuss the advantages and disadvantages of using MT for CLTR.
- ISBN 13 : 1581122675
- ISBN 10 : 9781581122671
- Judul : A Machine Translation Approach to Cross Language Text Retrieval
- Pengarang : María Gabriela Fernandez-Diaz,
- Kategori : Language Arts & Disciplines
- Penerbit : Universal-Publishers
- Bahasa : en
- Tahun : 2005
- Halaman : 0
- Google Book : https://play.google.com/store/books/details?id=NPjXr_WrljYC&source=gbs_api
-
Ketersediaan :
This second part of the project proved essential for deciding the benefits that a
more complex linguistic based method can add to CLTR. For this part of the
project, I worked with TRECQ1-10. First of all, it is important to mention the fact
that the IR system was not helpful for the experimentation involved here. SIRE is
designed to deal with words individually, and thus a comparison between the
retrieval of the original TREC queries using strings of individual words as
opposed to other ...