Demo paper accepted at IJCNLP

The demo paper “MUSST: A Multilingual Syntactic Simplification Tool” authored by Carolina Scarton, Alessio Palmero Aprosio, Sara Tonelli, Tamara Martín Wanton and Lucia Specia has been accepted at the 8th International Joint Conference on Natural Language Processing (IJCNLP 2017) to be held in Taipei, Taiwan.

Abstract:

In this paper we describe MUSST, a multilingual syntactic simplification tool. The tool supports sentence simplifications for English, Italian and Spanish, but it can be extended to other languages. Our implementation includes a set of general-purpose simplification rules, as well as a sentence selection module (to select sentences to be simplified) and a confidence model (to select only correct simplifications). The tool was implemented in the context of the European project SIMPATICO on text simplification for Public Administrations (PA) documents. Our evaluation on a set of sentences in the PA domain shows that we obtain suitable simplifications for 76% of the cases in English, 80% of the cases in Spanish and 78% of the cases in Italian.