Normal view MARC view ISBD view

SynCro - parsing simple Croatian sentences / Vučković, Kristina ; Bekavac, Božo ; Dovedan, Zdravko.

By: Kocijan, Kristina.
Contributor(s): Bekavac, Božo [aut] | Dovedan Han, Zdravko [aut].
Material type: materialTypeLabelArticleDescription: 207-217.Other title: SynCro - Parsing Simple Croatian Sentences [Naslov na engleskom:].Subject(s): 5.04 | 6.03 | parse trees, partial parser, simple sentences, Croatian language, NooJ hrv | parse trees, partial parser, simple sentences, Croatian language, NooJ engOnline resources: Click here to access online In: NooJ 2009 International Conference and Workshop (08-10.06.2009. ; Touzeur, Tunis) Finite State Language Engineering: NooJ 2009 International Conference and Workshop str. 207-217Ben Hamadou, Abdelmajid ; Mesfar, Slim ; Silberztein, MaxSummary: In this paper, authors will present derivation of syntactic trees for simple active declarative Croatian sentences. The paper is a continuation of authors PhD thesis in which partial parser and NERC for Croatian language have been implemented and described. This is, to our knowledge, first attempt to build parser for Croatian using NooJ or Intex graphs. At this stage, only parse trees that show the following combinations, are described: subject - verb ; subject - verb - direct object ; subject - verb - direct (and, or) indirect object. In addition to all these combinations, we can detect adverbial phrases of place and of time for any given set and genitive <NP> compliments for all <NP>s in a sentence. Mostly free word order allowed in Croatian sentences presents quite a challenge for building comprehensive syntactic parse trees for them. But grammatical relations in Croatian sentences are usually encoded by a particular case. This fenomena enables easier detection of constituents, even if they are freely reordered (scrambled). The goal of prototype syntactic parser called SynCro is to reach high accuracy in terms of precision while recall is, as less important at this stage, left for further rule expansion. At the end of the paper, the results will be evaluated through precision to show the adequacy of the model.
Tags from this library: No tags from this library for this title. Log in to add tags.
No physical items for this record

In this paper, authors will present derivation of syntactic trees for simple active declarative Croatian sentences. The paper is a continuation of authors PhD thesis in which partial parser and NERC for Croatian language have been implemented and described. This is, to our knowledge, first attempt to build parser for Croatian using NooJ or Intex graphs. At this stage, only parse trees that show the following combinations, are described: subject - verb ; subject - verb - direct object ; subject - verb - direct (and, or) indirect object. In addition to all these combinations, we can detect adverbial phrases of place and of time for any given set and genitive compliments for all s in a sentence. Mostly free word order allowed in Croatian sentences presents quite a challenge for building comprehensive syntactic parse trees for them. But grammatical relations in Croatian sentences are usually encoded by a particular case. This fenomena enables easier detection of constituents, even if they are freely reordered (scrambled). The goal of prototype syntactic parser called SynCro is to reach high accuracy in terms of precision while recall is, as less important at this stage, left for further rule expansion. At the end of the paper, the results will be evaluated through precision to show the adequacy of the model.

Projekt MZOS 130-1300646-1002

Projekt MZOS 130-1300646-1776

ENG

There are no comments for this item.

Log in to your account to post a comment.

Powered by Koha