Matteo Grella

Followers

Following

Co-authors

Public Views

Uppsala University

Hacettepe University

University of Notre Dame

Evangelia Oikonomou

KU Leuven

Chen Yang

Hefei University of technology

Farhad Nooralahzadeh

University of Oslo

Edward Fox

Nagiza Samatova

InterestsView All (12)

Uploads

Papers by Matteo Grella

Non-Projective Dependency Parsing via Latent Heads Representation (LHR)

by Matteo Grella and Simone Cangialosi

In this paper we introduce a novel approach based on a bidirectional recurrent autoencoder to per... more In this paper we introduce a novel approach based on a bidirectional recurrent autoencoder to perform globally optimized non-projective dependency parsing via semi-supervised learning. The syntactic analysis is completed at the end of the neural process that generates a Latent Heads Representation (LHR), without any algorithmic constraint and with a linear complexity. The resulting " latent syntactic structure " can be used directly in other semantic tasks. The LHR is transformed into the usual dependency tree computing a simple vectors similarity. We believe that our model has the potential to compete with much more complex state-of-the-art parsing architectures.

Download

Taking Advantage of BiLSTM Encoding To Handle Punctuation In Dependency Parsing: A Brief Idea

by Matteo Grella and Matteo Grella

In the context of the bidirectional-LSTMs neural parser (Kiperwasser and Goldberg, 2016), an idea... more In the context of the bidirectional-LSTMs neural parser (Kiperwasser and Goldberg, 2016), an idea is proposed to initialize the parsing state without punctuation-tokens but using them for the BiLSTM sentence encoding. The relevant information brought by the punctuation-tokens should be implicitly learned using the errors of the recurrent contributions only.

Download

Experiments with a Constraint-based Dependency Parser

This brief article describes our contribution to the Evalita 2011 Dependency Parsing Task. The It... more This brief article describes our contribution to the Evalita 2011 Dependency Parsing Task. The Italian grammar has been expressed for the first time as a set of constraints that need to be satisfied by any parse tree. The constraints propagation technique is then applied to restrict possible analyses. Multiple solutions of a given sentence have been reduced to one (structural disambiguation) by weighting each relation of each different solution according to the number of occurrences of that relation in the indexed version of Italian Wikipedia created for the purpose. A detailed analysis of the results is given, including some consideration on the difference between the LAS and UAS values. The attachment score obtained is 96.16%, giving the best result so far for a dependency parser for the Italian language.

Download

Rule-based creation of timeML documents from dependency trees

by Matteo Grella and Tommaso Caselli

AI* IA 2011: Artificial Intelligence Around …, Jan 1, 2011

The access to information through content has become the new frontier in NLP. Innovative annotati... more