Identifying synonymous expressions from a bilingual corpus for example-based machine translation

Abstract

Example-based machine translation (EBMT) is based on a bilingual corpus. In EBMT, sentences similar to an input sentence are retrieved from a bilingual corpus and then output is generated from translations of similar sentences. Therefore, a similarity measure between the input sentence and each sentence in the bilingual corpus is important for EBMT. If some similar sentences are missed from retrieval, the quality of translations drops. In this paper, we describe a method to acquire synonymous expressions from a bilingual corpus and utilize them to expand retrieval of similar sentences. Synonymous expressions are acquired from dierences in synonymous sentences. Synonymous sentences are clustered by the equivalence of translations. Our method has the advantage of not relying on rich linguistic knowledge, such as sentence structure and dictionaries. We demonstrate the eect on applying our method to a simple EBMT.

Acquiring Synonyms from Monolingual Comparable Texts

“…Shimohata et al [13] extracted lexical paraphrases based on the substitution operation of edit operations. Text pairs having more than three edit distances are excluded from extraction.…”

A corpus-centered approach to spoken language translation

“…(2) Finch et al (2002) cluster sentences in a handcrafted paraphrase corpus (Sugaya et al, 2002) to obtain pairs that are similar to each other for training SMT models, then by using the models the decoder generates a paraphrase. The experimental results indicate that (i) the EBMT based on normalization had increased coverage (Shimohata et al, 2002b) and (ii) the SMT created on the normalized sentences had a reduced word-error-rate (Watanabe et al, 2002a). Imamura et al (2003) proposed a calculation that measures the literalness of a translation pair and called it TCR.…”