Automatic Target Word Disambiguation Using Syntactic Relationships

Automatic Target Word Disambiguation Using Syntactic Relationships Ebony Domingo1 and Rachel Edita Roxas2 1 Computer Science Department, College of Science and Information Technology, Ateneo de Zamboanga, La Purisima St., Zamboanga, Philippines 7000 2 Software Technology Department, Dela Salle University-Manila, 2401 Taft Avenue, Malate, Manila, Philippines 1000 {domingoeboc@yahoo.com, roxasr@dlsu.edu.ph} Abstract. Multiple target translations are due to several meanings of source words, and various target word equivalents depending on the context of the source word. Thus, an automated approach is presented for resolving target-word selection, based on “word-to-sense” and “sense-to-word” source-translation relationships, using syntactic relationships (subject-verb, verb-object, adjective- noun). Translation selection proceeds from sense disambiguation of source words based on knowledge from a bilingual dictionary with sense profiles and word similarity measures from WordNet, and selection of a target word using statistics from a target corpus. Test results using English to Tagalog translations showed an overall 64% accuracy for selecting word translation with a standardized precision of at least 80% for generating expected translations using 200 sentences with ambiguous words (an average of 4 senses) in three categories: nouns, verbs, and adjectives, using 145,746 word pairs in syntactic relationships, extracted from target corpora (317,113 words). Keywords: word sense disambiguation, machine translation. 1 Introduction Target word disambiguation is a task in machine translation where a decision has to be made on which of a set of alternative target-language words is the most appropriate translation of a source-language word [1], a process familiarly known as translation selection. For instance, the correct translation of the word ‘wash’ in Tagalog could be hilamos, hugas, laba, etc. depending on the object noun of the source verb ‘wash’. Several methods have been developed for target-word disambiguation on different types of corpora using different nature of word translations. Techniques exploit monolingual corpora on either the target language (target language based) or the source language to resolve lexical ambiguities. Target language based approaches include the use of statistics on lexical relations [3], estimation of translation probability using a language model of the target language [5], [1]. Other methods exploit information from the source language for disambiguation such as distributional clustering [6]. A more recent and novel approach in translation selection is the hybrid method [2] based on the “word-to-sense and sense-to-word” relationship between source word and its translations, the method selects translation through two levels: sense disambiguation of a source word and selection of a target word. Other techniques worth mentioning in this field is the use of dependency triples on an unrelated monolingual corpus to select among translations of a given verb [4]. A more recent approach exploits content-aligned bilingual corpora for phrasal translations based on monolingual similarity and translation confidence of aligned phrases of two languages [7]. This research addresses resolving word translation ambiguity based on the idea of “word-to-sense and sense-to-word” relationship between source word and its translation using a bilingual dictionary and syntactic relations (subject-verb, verb-objects and adjective noun) on un-tagged, monolingual corpora in the target language with word sense disambiguation on source words. 374 2 …

Content truncated to 500 words