Learning to lemmatise Polish noun phrases

Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics, pages 701–709, Sofia, Bulgaria, August 4-9 2013. c⃝2013 Association for Computational Linguistics Learning to lemmatise Polish noun phrases Adam Radziszewski Institute of Informatics, Wrocław University of Technology Wybrze˙ze Wyspia´nskiego 27 Wrocław, Poland adam.radziszewski@pwr.wroc.pl Abstract We present a novel approach to noun phrase lemmatisation where the main phase is cast as a tagging problem. The idea draws on the observation that the lemmatisation of almost all Polish noun phrases may be decomposed into trans- formation of singular words (tokens) that make up each phrase. We perform eval- uation, which shows results similar to those obtained earlier by a rule-based sys- tem, while our approach allows to separate chunking from lemmatisation. 1 Introduction Lemmatisation of word forms is the task of find- ing base forms (lemmas) for each token in running text. Typically, it is performed along POS tagging and is considered crucial for many NLP applica- tions. Similar task may be defined for whole noun phrases (Degórski, 2011). By lemmatisation of noun phrases (NPs) we will understand assigning each NP a grammatically correct NP correspond- ing to the same phrase that could stand as a dic- tionary entry. The task of NP lemmatisation is rarely con- sidered, although it carries great practical value. For instance, any keyword extraction system that works for a morphologically rich language must deal with lemmatisation of NPs. This is because keywords are often longer phrases (Turney, 2000), while the user would be confused to see inflected forms as system output. Similar situation happens when attempting at terminology extraction from domain corpora: it is usually assumed that do- main terms are subclass of NPs (Marciniak and Mykowiecka, 2013). In (1) we give an example Polish noun phrase (‘the main city of the municipality’). Through- out the paper we assume the usage of the tagset of the National Corpus of Polish (Przepiórkowski, 2009), henceforth called NCP in short. The or- thographic form (1a) appears in instrumental case, singular. Phrase lemma is given as (1b). Lem- matisation of this phrase consists in reverting case value of the main noun (miasto) as well as its adjective modifier (główne) to nominative (nom). Each form in the example is in singular number (sg), miasto has neuter gender (n), gmina is fem- inine (f). (1) a. głównym main inst:sg:n miastem city inst:sg:n gminy municipality gen:sg:f b. główne main nom:sg:n miasto city nom:sg:n gminy municipality gen:sg:f According to the lemmatisation principles ac- companying the NCP tagset, adjectives are lem- matised as masculine forms (główny), hence it is not sufficient to take word-level lemma nor the or- thographic form to obtain phrase lemmatisation. Degórski (2011) discuses some similar cases. He also notes that this is not an easy task and lemma of a whole NP is rarely a concatenation of lem- mas of phrase components. It is worth stressing that even the task of word-level lemmatisation is non-trivial for inflectional languages due to a large number of inflected forms and even larger num- …

Content truncated to 500 words