Authors:
 Xavier Lluís

 

References:
 http://nlp.lsi.upc.edu/jointparser/demo/index.php

 

Description:
 Jointparser is a data-driven parser that jointly performs both syntactic dependency parsing and shallow semantic parsing. The system is based on an extension of the Eisner algorithm and uses an online averaged preceptron as a learning method. Shallow semantic parsing is performed for nominal and verbal predicates. The system was presented in the context of the CoNLL-2008 shared task.

 

Functionality:

 

 Noun Phrase and Verbal Phrase identification, joint sintactic and semantic analysis (on-line for english sentences)

 

Technology:

 

 C++, web interface

 

Technical Requirements:

 

 Included svmlight (svmlight.joachims.org).

 

Modules:

 

Innovation:

 

 It was one of the two novel joint syntactic-semantic parsers presented at the CoNLL-2008 shared task. It achieved a reasonable performance given it is a built-from-scratch system.

 

Development:

 

 Xavier Lluís master's thesis (UPC 8/9/2008).

 

Publications:

 

  • Xavier Lluís and Lluís Márquez, A Joint Model for Parsing Syntactic and Semantic Dependencies, Proceedings of CoNLL-2008, 2008.

 

  • Xavier Lluís, advisor: Lluís Márquez, Joint Learning of Syntactic and Semantic Dependencies, Master's thesis, Technical University of Catalonia, 2008.

Contact: xlluis@lsi.upc.edu

Scroll to Top