DerivBase.Ru: a Derivational Morphology Resource for Russian

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 3937–3943 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 3937 DerivBase.Ru: a Derivational Morphology Resource for Russian Daniil Vodolazsky Novosibirsk State University Novosibirsk, Russia daniil.vodolazsky@mail.ru Abstract Russian morphology has been studied for decades, but there is still no large high coverage resource that contains the derivational families (groups of words that share the same root) of Russian words. The number of words used in different areas of the language grows rapidly, thus the human-made dictionaries published long time ago cannot cover the neologisms and the domain-specific lexicons. To fill such resource gap, we have developed a rule-based framework for deriving words and we applied it to build a derivational morphology resource named DerivBase.Ru, which we introduce in this paper. Keywords: derivational morphology, lexical resources, word graphs 1. Introduction Morphological derivation is a process of producing new words. E.g., a word worker is derived from a word work with derivational suffix -er. In many NLP tasks, one of- ten needs to find the words that are derivationally related (i.e. words that share the same root), for instance in para- phrase or plagiarism detection. This can be done with a derivational morphology resource. For a widely used Rus- sian language such resources exist1, but they are hand-made and limited with small number of words, and thus cannot be easily applied to the texts with new terminology. Tikhonov (1985) dictionary contains 145 000 words, but it is not available online, and its shorter version (Tikhonov, 2017) is not convenient for computational purposes. Our work is inspired by Zeller et al. (2013) and extends the proposed approach to Russian language. DErivBase is a large-coverage derivational lexicon for German which con- sists of derivational families, groups of lemmas which are derivationally related among each other. Russian is a fusional language, and it therefore presents a lot of prefixes and suffixes (both inflectional and deriva- tional). The most productive ways of word formation in Russian are suffixation: mechta (a dream) → mechta- tel’ (a dreamer), prefixation: krichat’ (to scream) → za- krichat’ (to start screaming), circumfixation: slepoy (blind) → [igrat’] vslepuyu ([playing] blind), compounding: pyl’ (dust), sosat’ (to suck) → pylesos (a vacuum cleaner), and abbreviation: matematicheskiy fakultet (mathematical de- partment) → matfak. In this work, we are focusing only on affixation, though our approach can be extended to other word formation types. Given that not all the possible outputs of a derivational rule are realised in the language, we additionally use informa- tion about an inflectional type (Zaliznyak, 1980) of a source word (for nouns, adjectives and verbs), as it helps us to de- sign rules more accurately. Every inflectional type repre- sents a special pattern that is common to all words of a given class. E. g., consider a verb um’e-t’ (can) and its wordforms um’e-ju (1per, sg), um’e-jet (3per, sg). A verb dela-t’ (do) — d’ela-ju — d’ela-jet conjugates the same way, therefore it belongs …

Content truncated to 500 words