Statistical Machine Translation Using Monolingual Corpora

METIS

Statistical Machine Translation Using Monolingual Corpora

The aim of the project was to assess the possibility to obtain free text translations of reasonably high quality from large annotated monolingual corpora with pattern-matching techniques.

The source text is tagged and lemmatised. Then, by using bilingual lexica and SL-TL tag mapping rules, a first output is obtained and fed to the pattern matching algorithms. The algorithms compute the similarity of the input to the annotated text in the monolingual corpus and return those strings that are minimally different from the input ones.

The basic idea of the project was investigated further within the project METIS II.

Κατάσταση
Completed
Έναρξη
Λήξη
Επιστημονικός Υπεύθυνος
Stella Markantonatou