Integration of Data from a Syntactic Lexicon into Generative and Discriminative Probabilistic Parsers

Proceedings of Recent Advances in Natural Language Processing, pages 363–370, Hissar, Bulgaria, 12-14 September 2011. Integration of Data from a Syntactic Lexicon into Generative and Discriminative Probabilistic Parsers Anthony Sigogne Universit´e Paris-Est, LIGM sigogne@univ-mlv.fr Matthieu Constant Universit´e Paris-Est, LIGM mconstan@univ-mlv.fr ´Eric Laporte Universit´e Paris-Est, LIGM laporte@univ-mlv.fr Abstract This article evaluates the integration of data extracted from a syntactic lexi- con, namely the Lexicon-Grammar, into several probabilistic parsers for French. We show that by modifying the Part-of- Speech tags of verbs and verbal nouns of a treebank, we obtain accurate perfor- mances with a parser based on Probabilis- tic Context-Free Grammars (Petrov et al., 2006) and a discriminative parser based on a reranking algorithm (Charniak and John- son, 2005). 1 Introduction Syntactic lexicons are rich language resources that may contain useful data for parsers like sub- categorisation frames, as they provide, for each lexical entry, information about its syntactic be- haviors. Most of the time, these lexicons only deal with verbs. Few, like the Lexicon-Grammar (Gross, 1994), deal with other categories like nouns, adjectives or adverbs. Many works on symbolic parsing studied the use of a syntactic lexicon, in particular linguistic formalisms like Lexical-Functional Grammars [LFG] (Kaplan and Maxwell, 1994; Riezler et al., 2002; Sagot, 2006) or Tree Adjoining Grammars [TAG] (Joshi, 1987; Sagot and Tolone, 2009; de La Clergerie, 2010). For probabilistic parsing, we can cite LFG (Cahill, 2004; O’Donovan et al., 2005; Schluter and Gen- abith, 2008), Head-Driven Phrase Structure Gram- mar [HPSG] (Carroll and Fang, 2004) and Proba- bilistic Context-Free Grammars [PCFG] (Briscoe and Carroll, 1997; Deoskar, 2008). The latter has incorporated valence features to PCFGs and lex- icons and observes slight improvements on per- formances. However, lexical resources that con- tain valence features were obtained automatically from a corpus. Furthermore, valence features are mainly used on verbs. In this paper, we will show how we can exploit information contained in the Lexicon-Grammar in order to improve probabilis- tic parsers. We will in particular focus on verbs and verbal nouns1. In section 2, we describe the probabilistic parsers used in our experiments. Section 3 briefly intro- duces the Lexicon-Grammar. We detail informa- tion contained in this lexicon that can be used for parsing. Then, in section 4, we present methods to integrate this information into parsers and, in sec- tion 5, we describe our experiments and discuss the obtained results. 2 Statistical parsers In our experiments, we used two types of parsers: a generative parser that generates the n-best parses (n most probable parses) for a sentence accord- ing to a PCFG; a reranker that reranks the n-best parses generated from the PCFG parser according to a discriminative probabilistic model. 2.1 Non-lexicalized PCFG parser The PCFG parser, used into our experiments, is the Berkeley Parser (called BKY thereafter) (Petrov et al., 2006)2. This parser is based on a non- lexicalized PCFG model. The main problem of non-lexicalized context-free grammars is that pre- terminal symbols encode too general information which weakly discriminates syntactic ambiguities. BKY tries to handle the …

Content truncated to 500 words