Andrea Gesmundo
Biography
I am a PhD student in Computer Science working on the COMTIS Project at the Université de Genève
and a member of the the Computational Learning and Computational Linguistics Research Group (CLCL).
I am advised by Paola Merlo and James Henderson. My research interests include the application of Machine Learning Techniques as Bayesian Models, HMM, Perceptron Algorithm, to solve a wide range of Statistical Natural Language Processing tasks as Syntactic Parsing, Semantic Role Labeling, Part of Speech Tagging, Named Entity Recognition, or Shallow Parsing, and more recently Machine Translation.
You can see my CV here.
Publications
Heuristic Cube Pruning in Linear Time
( previously known as: Approximate Cube Pruning in Linear Time )
Andrea Gesmundo, Giorgio Satta, James Henderson. ACL (2012)
bibtex
Lemmatisation as a Tagging Task
Andrea Gesmundo, Tanja Samardzic. ACL (2012)
NEW! Try this model Online: Online Lemmatiser
Code: BTagger@github
bibtex
HadoopPerceptron: a Toolkit for Distributed Perceptron Training and Prediction with MapReduce
Andrea Gesmundo, Nadi Tomeh. EACL (2012), Demo Session
NEW! version 2.0 released
Code: HadoopPerceptron@github
bibtex
(57% acceptance rate)
Heuristic Search for Non-Bottom-Up Tree Structure Prediction
Andrea Gesmundo, James Henderson. EMNLP (2011)
poster
bibtex
(24% acceptance rate)
Bidirectional Sequence Classification for Tagging Tasks with Guided Learning
Andrea Gesmundo. TALN (2011)
Code: BTagger@github
Try it on MLcomp: mlcomp.org/programs/966
bibtex
Faster Cube Pruning
Andrea Gesmundo, James Henderson. IWSLT (2010)
Faster Cube Pruning has been integrated in CDEC
Code: github.com/redpony/cdec
Cdec documentation and build instructions: cdec-decoder.org
Data: cdec-decoder.org/index.php?title=Cdec_sample_grammar_and_test_set
To switch on Algorithm 2 add the command line option "-I Fast_cube_pruning"
To switch on Algorithm 3 add the command line option "-I Fast_cube_pruning_2"
bibtex
Bidirectional Sequence Classification for Part of Speech Tagging
Andrea Gesmundo. EVALITA (2009)
(2nd result out of 11, 1st non-combined system)
Try this model Online: Italian POS Tagger
Code: github.com/agesmundo/BTagger
bibtex
Bidirectional Sequence Classification for Named Entities Recognition
Andrea Gesmundo. EVALITA (2009)
(2nd result out of 12, 1st non-combined system)
Code: github.com/agesmundo/BTagger
bibtex
A Latent Variable Model of Synchronous Syntactic-Semantic Parsing for Multiple Languages
Andrea Gesmundo, James Henderson, Paola Merlo and Ivan Titov. CoNLL (2009)
(3rd result out of 14, 1st result for syntax)
Code: github.com/agesmundo/IDParser
bibtex
More Code & Data
Minimum Bayes-Risk for Hierarchical Machine Translation
Language Model for Italian
Trigrams, from Europarl corpus v6 (~700k sentences).
Online Lemmatizer for Serbian and Croatian
Tutorials slides
Introduction to Sampling and Reversible-Jump Markov Chain Monte Carlo.
Introduction to Bayesian Models and Dirichlet Process.
Master's Thesis
Natural Language Processing with Bidirectional Features.
Andrea Gesmundo. Università di Padova (2007).
Thesis supervisor: Giorgio Satta
Contacts
email:
address:
Battelle A
7, route de Drize
1227 Carouge
SWITZERLAND
phone:
+41 22 37 90136
twitter:
@agesmundo
Google Scholar
