Building parallel corpora from the Web

Mgr. Jan Pomikálek, Ph.D.

Rigorózní práce

Anotace:

Parallel corpora are a valuable resource for many fields in computational linguistics, e.g. machine translation, cross language information retrieval (CLIR), lexicography. Unfortunately, the sources of parallel texts are very limited. On the other hand, there is World Wide Web with billions of Web pages, some of which are mutual translations. Though its potential for retrieving bilingual texts awaits further development, studies exist which suggest that the Web can be used as a source for building parallel corpora, at least for some language pairs. However, the procedure for locating parallel texts on the Web is far from trivial. In this work I give an overview of the existing methods and algorithms for mining multilingual textual resources from the Web. I point out the disadvantages and limitations of these methods and propose some original ideas on how to overcome these drawbacks.

Abstract:

Parallel corpora are a valuable resource for many fields in computational linguistics, e.g. machine translation, cross language information retrieval (CLIR), lexicography. Unfortunately, the sources of parallel texts are very limited. On the other hand, there is World Wide Web with billions of Web pages, some of which are mutual translations. Though its potential for retrieving bilingual texts awaits further development, studies exist which suggest that the Web can be used as a source for building parallel corpora, at least for some language pairs. However, the procedure for locating parallel texts on the Web is far from trivial. In this work I give an overview of the existing methods and algorithms for mining multilingual textual resources from the Web. I point out the disadvantages and limitations of these methods and propose some original ideas on how to overcome these drawbacks.

Keywords

corpus, text corpora, web-derived corpora, parallel corpora

Jazyk práce: angličtina

Datum vytvoření / odevzdání či podání práce: 17. 6. 2008

Obhajoba závěrečné práce

• Obhajoba proběhla 23. 6. 2008

Masarykova univerzita

Fakulta informatiky

Rigorózní řízení / obor:
Informatika / Informatika