PorSimples: simplification of Portuguese text for digital inclusion and accessibility

• Grant number: 07/54565-8
• Support Opportunities: Regular Research Grants
• Start date: November 01, 2007
• End date: April 30, 2010
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Agreement: Microsoft Research
• Principal Investigator: Sandra Maria Aluísio
• Grantee: Sandra Maria Aluísio
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

The main goal of PorSimples was to develop Natural Language Processing (NLP) technologies related to Text Adaptation (TA) to promote digital inclusion and accessibility for people with low levels of literacy. There are two general different approaches for TA: Text Simplification and Text Elaboration. The first can be defined as any task that reduces the lexical or syntactic complexity of a text, while trying to preserve meaning and information, and can be subdivided into Lexical and Syntactic Simplification, Automatic Summarization, and other techniques. Text Elaboration aims at clarifying and explaining information and making connections explicit in a text, for example, providing definitions or synonyms for words known to only a few speakers of a language. The technologies developed in PorSimples are available by means of three systems aimed to distinct users: 1) Figure 1. Main page of Simplifica an authoring system, called Simplifica, to help authors to produce simplified texts targeting people with low literacy levels (figure 1); 2) an assistive technology system, called Facilita, which explores the tasks of summarization and simplification to allow poor literate people to read Web content (figure 2), and 3) a web content adaptation tool, named Educational Facilita, for assisting low-literacy readers to perform detailed reading. It exhibits questions that clarify the semantic relations linking verbs to their arguments, highlights the associations amongst the main ideas of the texts and the named entities, and perform lexical elaboration. Currently, Educational Facilita only explores the NLP tasks of lexical elaboration and named entity labeling. (AU)