SB: mmSystem - Using Decompositional Semantics for Lexical Simplification

First Joint Conference on Lexical and Computational Semantics (*SEM), pages 482–486, Montr´eal, Canada, June 7-8, 2012. c⃝2012 Association for Computational Linguistics SB: mmSystem - Using Decompositional Semantics for Lexical Simplification Marilisa Amoia Department of Applied Linguistics University of Saarland m.amoia@mx.uni-saarland.de Massimo Romanelli DFKI GmBH Saarbrcken, Germany romanell@dfki.de Abstract In this paper, we describe the system we sub- mitted to the SemEval-2012 Lexical Simplifi- cation Task. Our system (mmSystem) com- bines word frequency with decompositional semantics criteria based on syntactic structure in order to rank candidate substitutes of lexical forms of arbitrary syntactic complexity (one- word, multi-word, etc.) in descending order of (cognitive) simplicity. We believe that the proposed approach might help to shed light on the interplay between linguistic features and lexical complexity in general. 1 Introduction Lexical simplification is a subtask of the more gen- eral text simplification task which attempts at re- ducing the cognitive complexity of a text so that it can be (better) understood by a larger audience. Text simplification has a wide range of applications which includes applications for the elderly, learners of a second language, children or people with cog- nitive deficiencies, etc. Works on text simplification mostly focus on re- ducing the syntactic complexity of the text (Sid- dharthan, 2011; Siddharthan, 2006) and only little work has addressed the issue of lexical simplifica- tion (Devlin, 1999; Carroll et al., 1999). The Lexical Simplification Task (Specia et al., 2012) proposed within the SemEval-2012 is the first attempt to explore the nature of the lexical simpli- fication more systematically. This task requires par- ticipating systems, given a context and a target word, to automatically generate a ranking of substitutes, i.e. lexical forms conveying similar meanings to the target word, such that cognitively simpler lexi- cal forms are ranked higher than more difficult ones. In this paper, we describe the system we sub- mitted to the SemEval-2012 Lexical Simplification Task. In order to rank the candidate substitutes of a lexical form in descending order of simplicity, our system (mmSystem) combines word frequency with decompositional semantics criteria based on syntac- tic structure. The mmSystem achieved an average ranking if compared with the other participating sys- tems and the baselines. We believe that the approach proposed in this paper might help to shed light on the interplay between linguistic features and cogni- tive complexity in general. 2 The Lexical Simplification Task The SemEval-2012 Lexical Simplification Task re- quires participating systems to automatically gen- erate a ranking of lexical forms conveying similar meanings on cognitive simplicity criteria and can be defined as follows. Given a short text C called the context and which generally corresponds to a sen- tence, a target word T and a list LS of candidate substitutes for T, i.e. a list of quasi-synonyms of the target word, the task for a system consists in pro- viding a ranking on LS such that the original list of substitutes is sorted over simplicity, from the cogni- tively simplest to the cognitively most difficult lexi- cal form. As …

Content truncated to 500 words