MERALI at SemEval-2017 Task 2 Subtask 1: a Cognitively Inspired approach

Proceedings of the 11th International Workshop on Semantic Evaluations (SemEval-2017), pages 245–249, Vancouver, Canada, August 3 - 4, 2017. c⃝2017 Association for Computational Linguistics MERALI at SemEval-2017 Task 2 Subtask 1: a Cognitively Inspired approach Enrico Mensa University of Turin, Italy Dipartimento di Informatica mensa@di.unito.it Daniele P. Radicioni University of Turin, Italy Dipartimento di Informatica radicion@di.unito.it Antonio Lieto University of Turin, Italy ICAR-CNR, Palermo, Italy lieto@di.unito.it Abstract In this paper we report on the participa- tion of the MERALI system to the Se- mEval Task 2 Subtask 1. The MER- ALI system approaches conceptual sim- ilarity through a simple, cognitively in- spired, heuristics; it builds on a linguistic resource, the TTCSE, that relies on Babel- Net, NASARI and ConceptNet. The lin- guistic resource in fact contains a novel mixture of common-sense and encyclope- dic knowledge. The obtained results point out that there is ample room for improve- ment, so that they are used to elaborate on present limitations and on future steps. 1 Introduction Defining conceptual representations along with their associated reasoning procedures required and still requires truly interdisciplinary efforts, in- volving psychologists (Miller and Charles, 1991; Barsalou, 1999; Malt et al., 2015), philoso- phers (Machery, 2009; G¨ardenfors, 2014), neu- roscientists (Vigliocco et al., 2014), and com- puter scientists (Resnik, 1998; Agirre et al., 2009; Pilehvar and Navigli, 2015). Today, the ever- growing number of applications of semantic tech- nologies demand for further investigation on con- cepts’ meaning: this fact explains the popularity of issues rooted in and related to conceptual similar- ity, and the success of the present Semantic Word Similarity task (Camacho-Collados et al., 2017). In this paper we present an approach to the computation of conceptual similarity based on a novel lexical resource, the TTCSE—so dubbed af- ter Terms to Conceptual Spaces-Extended— that has been acquired by integrating two different sorts of linguistic resources, such as the ency- clopedic knowledge available in BabelNet (Nav- igli and Ponzetto, 2012) and NASARI (Camacho- Collados et al., 2015), and the common-sense grasped by ConceptNet (Speer and Havasi, 2012). The resulting representation enjoys the interest- ing property of being anchored to both resources, thereby providing a uniform conceptual access grounded on the sense identifiers provided by Ba- belNet. The TTCSE provides a conceptual representa- tion inspired to Conceptual Spaces (CSs), a ge- ometric representation framework where knowl- edge is represented as a set of limited though cog- nitively relevant quality dimensions (G¨ardenfors, 2014). The CSs framework has been recently used to extend and complement the representa- tional and inferential power allowed by formal on- tologies with special emphasis on dealing with the corresponding typicality-based conceptual reason- ing (Lieto et al., 2015, 2017); in this setting, the TTCSE aims at providing a wide-coverage, cog- nitively based linguistic resource for this sort of knowledge, by extending previous work (Lieto et al., 2016; Mensa et al., 2017). 2 Concept Representation in the TTCSE Concepts representation in the TTCSE is consistent with CSs: each concept c is provided with a vec- tor representation ⃗c providing …

Content truncated to 500 words