UFRGS&LIF at SemEval-2016 Task 10: Rule-Based MWE Identification and Predominant-Supersense Tagging

Proceedings of SemEval-2016, pages 910–917, San Diego, California, June 16-17, 2016. c⃝2016 Association for Computational Linguistics UFRGS&LIF at SemEval-2016 Task 10: Rule-Based MWE Identification and Predominant-Supersense Tagging Silvio Ricardo Cordeiro1,2 and Carlos Ramisch2 and Aline Villavicencio1 1 Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil 2Aix Marseille Université, CNRS, LIF UMR 7279 {srcordeiro,avillavicencio}@inf.ufrgs.br carlos.ramisch@lif.univ-mrs.fr Abstract This paper presents our approach towards the SemEval-2016 Task 10 – Detecting Minimal Semantic Units and their Meanings. Systems are expected to provide a representation of lex- ical semantics by (1) segmenting tokens into words and multiword units and (2) providing a supersense tag for segments that function as nouns or verbs. Our pipeline rule-based sys- tem uses no external resources and was imple- mented using the mwetoolkit. First, we ex- tract and filter known MWEs from the train- ing corpus. Second, we group input tokens of the test corpus based on this lexicon, with spe- cial treatment for non-contiguous expressions. Third, we use an MWE-aware predominant- sense heuristic for supersense tagging. We ob- tain an F-score of 51.48% for MWE identifi- cation and 49.98% for supersense tagging. 1 Introduction Accurate segmentation and semantic disambigua- tion of minimal text units is a major challenge in the general pipeline of NLP applications. A ma- chine translation system, for example, needs to de- cide what is the intended meaning for a given word or phrase in its context, so that it may translate it into an equivalent meaning in the target language. While determining the meaning of single words is a difficult task on its own, the problem is com- pounded by the pervasiveness of Multiword Expres- sions (MWEs). MWEs are semantic units that span over multiple lexemes in the text (e.g. dry run, look up, fall flat). Their meaning cannot be inferred by applying regular composition rules on the meanings of their component words. The task of semantic tag- ging is thus deeply intertwined with the identifica- tion of multiword expressions. This paper presents our solution to the DiMSUM shared task (Schneider et al., 2016), where the evalu- ated systems are expected to perform both semantic tagging and multiword identification. Our pipeline system first detects and groups MWEs and then as- signs supersense tags, as two consecutive steps. For MWE identification, we use a task-specific instan- tiation of the mwetoolkit (Ramisch, 2015), han- dling both contiguous and non-contiguous MWEs with some degree of customization (Cordeiro et al., 2015). Additionally, MWE type-level candidates are extracted without losing track of their token- level occurrences, to guarantee that all the MWE occurrences learned from the training data are pro- jected onto the test corpus. For semantic tagging we adopted a predominant-sense heuristic. In the remainder of this paper, we present related work (§ 2), then we present and discuss the results of the MWE identification subsystem (§ 3) and of the supersense tagging subsystem (§ 4). We then con- clude and share ideas for future improvements (§ 5). 2 Related Work Practical solutions for rule-based MWE identifica- …

Content truncated to 500 words