Skip to Main content Skip to Navigation
Book sections

Experiments with a French-Romanian Word Alignment System.

Abstract : In this paper, we present several experiments with a hybrid French-Romanian word alignment system. We combine statistical methods and linguistic resources (heuristic rules using linguistic information, dictionaries). The experiments evaluate the influence of various linguistic resources on the overall system's results. We use GIZA++ as a baseline system and we iteratively add cognates, collocations and heuristic rules to correct some alignment errors. For our experiments, we use training and evaluation corpora extracted from a freely available legal parallel corpus (DGT-TM). The results show that linguistic resources significantly improve the quality of the alignment.
Document type :
Book sections
Complete list of metadata

https://hal.inria.fr/hal-00870251
Contributor : Amalia Todirascu Connect in order to contact the contributor
Submitted on : Sunday, October 6, 2013 - 5:04:29 PM
Last modification on : Thursday, February 27, 2020 - 2:16:02 PM

Identifiers

  • HAL Id : hal-00870251, version 1

Collections

Citation

Mirabela Navlea, Amalia Todirascu. Experiments with a French-Romanian Word Alignment System.. Dan Tufiș, Vasile Rus, Corina Forăscu. Towards Multilingual Europe 2020: A Romanian Perspective, Editura Academiei Române, pp 225-240, 2013, ISBN 978-973-27-2282-4. ⟨hal-00870251⟩

Share

Metrics

Record views

63