UCM-I: A Rule-based Syntactic Approach for Resolving the Scope of Negation

First Joint Conference on Lexical and Computational Semantics (*SEM), pages 282–287, Montr´eal, Canada, June 7-8, 2012. c⃝2012 Association for Computational Linguistics UCM-I: A Rule-based Syntactic Approach for Resolving the Scope of Negation Jorge Carrillo de Albornoz, Laura Plaza, Alberto D´ıaz and Miguel Ballesteros Universidad Complutense de Madrid C/ Prof. Jos´e Garc´ıa Santesmases, s/n 28040 Madrid (Spain) {jcalbornoz,lplazam,albertodiaz,miballes}@fdi.ucm.es Abstract This paper presents one of the two contribu- tions from the Universidad Complutense de Madrid to the *SEM Shared Task 2012 on Re- solving the Scope and Focus of Negation. We describe a rule-based system for detecting the presence of negations and delimitating their scope. It was initially intended for process- ing negation in opinionated texts, and has been adapted to fit the task requirements. It first detects negation cues using a list of explicit negation markers (such as not or nothing), and infers other implicit negations (such as affixal negations, e.g, undeniable or improper) by us- ing semantic information from WordNet con- cepts and relations. It next uses the informa- tion from the syntax tree of the sentence in which the negation arises to get a first approxi- mation to the negation scope, which is later re- fined using a set of post-processing rules that bound or expand such scope. 1 Introduction Detecting negation is important for many NLP tasks, as it may reverse the meaning of the text affected by it. In information extraction, for instance, it is obviously important to distinguish negated informa- tion from affirmative one (Kim and Park, 2006). It may also improve automatic indexing (Mutalik et al., 2001). In sentiment analysis, detecting and deal- ing with negation is critical, as it may change the polarity of a text (Wiegand et al., 2010). How- ever, research on negation has mainly focused on the biomedical domain, and addressed the problem of detecting if a medical term is negated or not (Chap- man et al., 2001), or the scope of different negation signals (Morante et al., 2008). During the last years, the importance of process- ing negation is gaining recognition by the NLP re- search community, as evidenced by the success of several initiatives such as the Negation and Spec- ulation in Natural Language Processing workshop (NeSp-NLP 2010)1 or the CoNLL-2010 Shared Task2, which aimed at identifying hedges and their scope in natural language texts. In spite of this, most of the approaches proposed so far deal with negation in a superficial manner. This paper describes our contribution to the *SEM Shared Task 2012 on Resolving the Scope and Focus of Negation. As its name suggests, the task aims at detecting the scope and focus of nega- tion, as a means of encouraging research in negation processing. In particular, we participate in Task 1: scope detection. For each negation in the text, the negation cue must be detected, and its scope marked. Moreover, the event or property that is negated must be recognized. A comprehensive description of the task may be found in (Morante and Blanco, 2012). For the …

Content truncated to 500 words