WordNet is a well-known English lexical database widely used in language-related research. However, it is time-consuming and labor-insensitive to manually compile such a dictionary for the languages of the world. In this paper, we propose an automatic method of translating WordNet into languages other than English. In our approach, we utilize the ontological relations in WordNet and relevant collocations to obtain appropriate translations for a given WordNet synset. The method involves automatically identifying monosemous lemmas of the synset, automatically acquiring relevant collocations, automatically translating relevant words and phrases, and automatically extracting translations of the synset. We evaluate the proposed method on a set of WordNet synsets based on the human judgment. The experimental results show that the proposed method significantly outperforms the baseline of the most frequent translations.
WordNet是一部廣為人知的英文語意字典,並且廣泛應用於自然語言處理相關研究。然而,使用人工的方式編撰其他不同語言的WordNet,需要耗費大量的時間和人力。於本論文,我們提出一個自動的方法,翻譯WordNet到英文以外的其他語言。此一方法,對於一個被給定的WordNet同義詞集(synset) 進行翻譯,我們使用WordNet蘊含的本體關係(ontological relation)與同義詞集中相關搭配詞資訊。本方法包涵了自動辨識同義詞集中,單詞義的詞形(monosemous lemmas) ,自動取得相關的搭配詞,自動翻譯相關的字詞,最後自動抽取同義詞集的適當翻譯。我們隨機選取WordNet的一組同義詞集進行實驗並進行人工驗證,實驗結果顯示,相較於直接將同義詞集之最常見翻譯作為結果的基準方法,我們提出的方法更能準確地產生正確翻譯。