## Supertagging and full parsing (2004)

Venue: | In: Proceedings of the Workshop on Tree Adjoining Grammar and Related Formalisms (TAG+7 |

Citations: | 11 - 3 self |

### BibTeX

@INPROCEEDINGS{Nasr04supertaggingand,

author = {Alexis Nasr and Owen Rambow},

title = {Supertagging and full parsing},

booktitle = {In: Proceedings of the Workshop on Tree Adjoining Grammar and Related Formalisms (TAG+7},

year = {2004}

}

### Years of Citing Articles

### OpenURL

### Abstract

We investigate an approach to parsing in which lexical information is used only in a first phase, supertagging, in which lexical syntactic properties are determined without building structure. In the second phase, the best parse tree is determined without using lexical information. We investigate different probabilistic models for adjunction, and we show that, assuming hypothetically perfect performance in the first phase, the error rate on dependency arc attachment can be reduced to 2.3 % using a full chart parser. This is an improvement of about 50% over previously reported results using a simple heuristic parser. 1

### Citations

735 | Accurate Unlexicalized Parsing
- Klein, Manning
- 2003
(Show Context)
Citation Context ...he performance of the bilexical parsers seems to be plateauing. In fact recent work has questioned to what extent bilexical parsers even profit from bilexical information that they use (Gildea, 2001; =-=Klein and Manning, 2003-=-). 2.2 But Is Supertag-Based Parsing Feasible? Bangalore and Joshi (1999) claim that supertagging is “almost parsing”. What this means is that the syntactic information provided by supertags is so ric... |

529 | Three generative, lexicalised models for statistical parsing
- Collins
- 1997
(Show Context)
Citation Context ...ncies – probabilities that one particular head depends on another – to guide the parser. Current parsers achieve an score of about 90% when measuring just the accuracy of choosing these dependencies (=-=Collins, 1997-=-; Chiang, 2000; Clark et al., 2002; Hockenmaier and Steedman, 2002). Interestingly, the choice of formalism (headed CFG, TAG, or CCG) does not greatly change the parsers’ accuracy, presumably because ... |

144 | Supertagging: An Approach to Almost Parsing
- Bangalore, Joshi
- 1999
(Show Context)
Citation Context ...hes, the underlying information is the same – word-word dependencies. Supertagging followed by “lightweight” parsing has been proposed as an alternative to full parsing. The idea behind supertagging (=-=Bangalore and Joshi, 1999-=-) is to extend the notion of “tag” from a part of speech or a part of speech including morphological information to a tag that represents rich syntactic information as well, in particular active valen... |

123 |
Stochastic lexicalized treeadjoining grammars
- Schabes
- 1992
(Show Context)
Citation Context ...up an analysis. The general model is therefore: P (A) = P (Root) � × × × A∈A|O(A)=subst � s∈S,i∈nodes(s) � s∈S,i∈nodes(s) P (A) P (LEF T (s, i)) P (RIGHT (s, i)) This basically follows (Resnik, 1992; =-=Schabes, 1992-=-). The models we discuss here differ in how to compute the terms P (RIGHT (s, i)) and P (LEF T (s, i)). The probability of each attachment is estimated by maximum likelihood (the counts are obtained i... |

111 | An Alternative Conception of TreeAdjoining Derivation - Schabes, Shieber - 1994 |

101 | Corpus variation and parser performance
- Gildea
- 2001
(Show Context)
Citation Context ...specially as the performance of the bilexical parsers seems to be plateauing. In fact recent work has questioned to what extent bilexical parsers even profit from bilexical information that they use (=-=Gildea, 2001-=-; Klein and Manning, 2003). 2.2 But Is Supertag-Based Parsing Feasible? Bangalore and Joshi (1999) claim that supertagging is “almost parsing”. What this means is that the syntactic information provid... |

88 |
Probabilistic Tree-Adjoining Grammar as a Framework for Statistical Natural Language
- Resnik
- 1992
(Show Context)
Citation Context ...nts that make up an analysis. The general model is therefore: P (A) = P (Root) � × × × A∈A|O(A)=subst � s∈S,i∈nodes(s) � s∈S,i∈nodes(s) P (A) P (LEF T (s, i)) P (RIGHT (s, i)) This basically follows (=-=Resnik, 1992-=-; Schabes, 1992). The models we discuss here differ in how to compute the terms P (RIGHT (s, i)) and P (LEF T (s, i)). The probability of each attachment is estimated by maximum likelihood (the counts... |

87 | Data and Models for Statistical Parsing with Combinatory Categorial Grammar
- Hockenmaier
- 2003
(Show Context)
Citation Context ...pends on another – to guide the parser. Current parsers achieve an score of about 90% when measuring just the accuracy of choosing these dependencies (Collins, 1997; Chiang, 2000; Clark et al., 2002; =-=Hockenmaier and Steedman, 2002-=-). Interestingly, the choice of formalism (headed CFG, TAG, or CCG) does not greatly change the parsers’ accuracy, presumably because in all approaches, the underlying information is the same – word-w... |

80 | Tree Insertion Grammar: A CubicTime Parsable Formalism That Lexicalizes Context-Free Grammar Without Changing the Trees Produced - Schabes, Waters - 1994 |

54 | Bilexical Grammars and Their Cubic-Time Parsing Algorithms
- Eisner
- 2000
(Show Context)
Citation Context ...ord to another, and there are fewer items. Furthermore, our FSMs are built left-to-right, while Evans and Weir only explore FSMs constructed bottom-up from the lexical anchor of the tree (not unlike (=-=Eisner, 2000-=-)). As a result, we can perform a strict left-to-right parse, which is not straightforwardly possible in standard TAG parsing using FSMs. Our parsing algorithm is similar to the work of Alshawi et al.... |

32 | A uniform method of grammar extraction and its applications
- Xia, Palmer, et al.
- 2000
(Show Context)
Citation Context ...onvert it to an RTN. This first step does not represent the research reported in this paper, and we describe it only for the sake of clarity. We use the approach of (Chen, 2001) (which is similar to (=-=Xia et al., 2000-=-) and (Chiang, 2000)). We use sections 02 to 21 of the Penn Treebank. However, we optimize the head percolation in the grammar extraction module to create meaningful dependency structures, rather than... |

23 | Capturing cfls with Tree Adjoining Grammars - Rogers - 1994 |

11 | Automaton-based parsing for lexicalized grammars - Evans, Weir - 1997 |

4 |
Creating a finitestate parser with application semantics
- Rambow, Bangalore, et al.
- 2002
(Show Context)
Citation Context ...ad. His formalism is very close to the one we present, but we use a single automaton. Also, the relation to an independently proposed syntactic formalism such as TAG is less obvious. In related work (=-=Rambow et al., 2002-=-), we have used the same automata constructed from an extracted TAG for parsing, but instead of using them in a chart parser, we have used them to construct a single large FSM that produces a dependen... |