A Parallel WordNet for English, Swedish and Bulgarian

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 3008–3015 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 3008 A Parallel WordNet for English, Swedish and Bulgarian Krasimir Angelov Chalmers University, University of Gothenburg Gothenburg, Sweden krasimir@chalmers.se Abstract We present the parallel creation of a WordNet resource for Swedish and Bulgarian which is tightly aligned with the Princeton WordNet. The alignment is not only on the synset level, but also on word level, by matching words with their closest translations in each language. We argue that the tighter alignment is essential in machine translation and natural language generation. About one-fifth of the lexical entries are also linked to the corresponding Wikipedia articles. In addition to the traditional semantic relations in WordNet, we also integrate morphological and morpho-syntactic information. The resource comes with a corpus where examples from Princeton WordNet are translated to Swedish and Bulgarian. The examples are aligned on word and phrase level. The new resource is open-source and in its development we used only existing open-source resources. Keywords: WordNet, Translation, Morphology, Sense-annotated Corpora, Wikipedia 1. Introduction WordNet (Fellbaum, 1998) started as an effort to enumerate the senses of all English words, and to organize those senses into a network of interrelated senses. Soon after, the ef- fort was taken into other languages. First the EuroWordNet (Bloksma et al., 1996) for Dutch, Italian, Spanish, French, German, Czech and Estonian, and later the BalkaNet (Sta- mou et al., 2019) for Bulgarian, Czech, Greek, Romanian, Serbian and Turkish. There is also a multitude of projects for other languages. Most publicly available WordNet re- sources are aggregated by the Open Multilingual WordNet project (Bond and Paik, 2012). Initially not all of the resources were released with free ac- cess, but luckily some of them were made free later or are being replaced with alternative free resources. Our con- tribution is a free WordNet-like resource for Swedish and Bulgarian. The availability of prior resources is discussed in Section 2. We have, in addition, the goal that the new resource should be usable for applications in GF (Grammatical Framework; Ranta (2011)). GF is a programming language for natural language applications, whose primary focus is on multilin- guality. Any GF application makes use of a grammar and a lexicon. It is quite common for an application to be avail- able in five to ten languages simultaneously. Therefore, to minimize the effort for new applications, the framework of- fers a Resource Grammars Library (RGL) for many lan- guages (Ranta, 2009). The lexicon, however, is typically built from scratch, since it must be highly multilingual and application specific. These kinds of lexicons are hard to get. For every language they must contain a full-form inflection table, and the words that are translations of each other must be linked together through an interlingual index. We see the new resource as a lexical library which must be compatible with the already existing resource grammars. From there, the different applications can …

Content truncated to 500 words