iRead4Skills Heads to EMNLP 2025 with Intelligent Complexity Analyzer
The iRead4Skills project has reached an important milestone: its latest paper has been accepted for presentation at EMNLP 2025 System Demonstrations, one of the leading international conferences in natural language processing. In November, team members Mario Izquierdo Álvarez and Wafa AISSA will travel to Suzhou, China, to showcase the work.
The accepted paper introduces the iRead4Skills Intelligent Complexity Analyzer, an open-access platform developed to support educators and content creators working with low-literacy adults. The tool assesses the complexity of written texts and provides diagnostic feedback to guide the adaptation of learning materials.
Key features of the system include:
- Multilingual support in Portuguese, Spanish, and French.
- Automatic classification of texts into four difficulty levels using state-of-the-art NLP models.
- Diagnostic evaluation of texts across four dimensions: textual structure, lexicon, syntax, and semantics.
By offering feedback at multiple levels of granularity, the platform helps teachers and developers better tailor materials to learners’ needs.
The development of the Intelligent Complexity Analyzer is the result of extensive teamwork across the iRead4Skills consortium. Alongside the presenting authors, the paper’s co-authors include Raquel Amaro, David Antunes, Thibault Bañeras-Roux, Jorge Baptista, Alejandro Catalá, Luís Correia, Thomas François, Marcos Garcia González, Nuno Mamede, Vasco Nuno Martins, Miguel Neves, Eugénio Ribeiro, Sandra Rodríguez Rey, and Elodie Vanzeveren.
The upcoming presentation at EMNLP provides an opportunity not only to showcase the iRead4Skills tool but also to engage with the global NLP community, exchange insights, and gather feedback to continue advancing the project’s mission: fostering innovation in adult literacy.
