VerbLexPor: a lexical resource with semantic roles for Portuguese

VerbLexPor: A Lexical Resource with Semantic Roles for Portuguese Leonardo Zilio1, Maria Jos´e Bocorny Finatto2, Aline Villavicencio1 1Institute of Informatics, Federal University of Rio Grande do Sul (Brazil), 2Institute of Linguistics, Federal University of Rio Grande do Sul (Brazil), Avenida Bento Gonc¸alves, 9500 - Agronomia, Porto Alegre - RS, Brazil, 91509-900 lzilio@inf.ufrgs.br, mfinatto@terra.com.br, avillavicencio@inf.ufrgs.br Abstract This paper presents a lexical resource developed for Portuguese. The resource contains sentences annotated with semantic roles. The sentences were extracted from two domains: Cardiology research papers and newspaper articles. Both corpora were analyzed with the PALAVRAS parser and subsequently processed with a subcategorization frames extractor, so that each sentence that contained at least one main verb was stored in a database together with its syntactic organization. The annotation was manually carried out by a linguist using an annotation interface. Both the annotated and non-annotated data were exported to an XML format, which is readily available for download. The reason behind exporting non-annotated data is that there is syntactic information collected from the parser annotation in the non-annotated data, and this could be useful for other researchers. The sentences from both corpora were annotated separately, so that it is possible to access sentences either from the Cardiology or from the newspaper corpus. The full resource presents more than seven thousand semantically annotated sentences, containing 192 different verbs and more than 15 thousand individual arguments and adjuncts. Keywords: Semantic Role Labeling, Lexical Resource, Corpus 1. Introduction Lexical resources with semantic role information are of great use in NLP applications, for example, in anaphora res- olution (Kong and Zhou, 2012), automatic summarization (Yoshikawa et al., 2012) and automatic translation (Feng et al., 2012; Jones et al., 2012). Semantic roles are labels that describe the semantic function of an argument in relation to a determined word class (usually, a verb). For instance, in Sentence A, we have a simple clause, in which the sub- ject (Roger) carried out the action (represented by the verb), while the direct object (the vase) is being modified by the event. Hence, we have the semantic roles of AGENT and PA- TIENT. On the other hand, in Sentence B, although we still have a subject (the vase), it is no longer the performer of the action, but the one who suffers a change of state (from being in one piece to being in more than one piece), so it plays the role of PATIENT. • Sentence A: [Roger] broke [the vase]. • Sentence B: [The vase] broke. In this simple example, we can see that semantic roles are closely related to syntax, but also depends on the semantic of the word it relates to (in this case, the verb). For Portuguese, there are only a few initiatives that were dedicated to developing resource with this type of semantic information: PropBank.Br (Duran et al., 2011; Duran and Alu´ısio, 2012), VerbNet.Br (Scarton, 2013) and FrameNet Brasil (Salom˜ao, 2009) are the most prominent, and all of them were directly derived from their English counterparts, as their …

Content truncated to 500 words