Parallel Corpora for WordNet Construction: Machine Translation vs. Automatic Sense Tagging

Abstract

Abstract. In this paper we present a methodology for WordNet construction based on the exploitation of parallel corpora with semantic annotation of the English source text. We are using this methodology for the enlargement of the Spanish and Catalan versions of WordNet 3.0, but the methodology can also be used for other languages. As big parallel corpora with semantic annotation are not usually available, we explore two strategies to overcome this problem: to use monolingual sense tagged corpora and machine translation, on the one hand; and to use parallel corpora and automatic sense tagging on the source text, on the other. With these resources, the problem of acquiring a WordNet from parallel corpora can be seen as a word alignment task. Fortunately, this task is well known, and some aligning algorithms are freely available.

Research on the semantic knowledge base in Chinese-English machine translation

“…It is usually used for the words translation such as in Ref. [10]. When translating the sentences, the results are usually blur, shift.…”

Automatic Wordnet Development for Low-Resource Languages using Cross-Lingual WSD

“…However, this kind of corpus is not easily available in most languages. In the research, which was conducted by Oliver and Climent (2012), two strategies for the automatic construction of these corpora are proposed: (i) by the machine translation of sense-tagged corpora, and (ii) by the automatic sense tagging of bi-lingual word-aligned corpora. The results for Spanish language showed that the first strategy works better than the second.…”

“…These methods are usually divided into merge and expansion approaches (Fellbaum & Vossen, 2012;Oliver & Climent, 2012;Erjavec & Fišer, 2006). However, there are methods that combine the merge and expansion models and benefit from the advantages of both approaches (Prabhu, Desai, Redkar, Prabhugaonkar, Nagvenkar, & Karmali, 2012;Apidianaki & Sagot, 2014).…”

“…On the other hand, methods that follow the expansion approach usually adopt WordNet structure and find the correct translation of the associated words with the WordNet synsets in the target language. In this process, multilingual resources such as comparable corpora (Kaji & Watanabe, 2006), parallel corpora (Oliver & Climent, 2012;Kazakov & Shahid, 2009;Fišer, 2009;Diab, 2004), thesaurus (Gunawan & Saputra, 2010), machine translators (Saveski & Trajkovski, 2010) and multiple bi-lingual machine readable dictionaries (Atserias, Climent, Farreres, Rigau, & Rodríguez, 2000;Patanakul & Charnyote, 2005;Bond, Isahara, Kanzaki, & Uchimoto, 2008;Lam, Al Tarouti, & Kalita, 2014) are used, which causes a bottleneck for low-resource languages.…”

Automatically constructing Wordnet Synsets

“…Wordnets, except the PWN, have been usually constructed by one of two approaches. The first approach translates the PWN to T (Bilgin et al, 2004), (Barbu and Mititelu, 2005), (Kaji and Watanabe, 2006), (Sagot and Fišer, 2008), (Saveski and Trajkovsk, 2010) and (Oliver and Climent, 2012); while the second approach builds a Wordnet in T, and then aligns it with the PWN by generating translations (Gunawan and Saputra, 2010). In terms of popularity, the first approach dominates over the second approach.…”