DISCALCULAI: A Computational Approach for the Identification of Dyscalculia in Educational Contexts

• Jean Alex Firmino Pinhata FATEC Ourinhos
• Mateus Natan Roes Salgueiro FATEC Ourinhos
• Alex Marino Gonçalves de Almeida FATEC Ourinhos
• Rosemeiry de Castro Prado FATEC Ourinhos

DOI: https://doi.org/10.5753/eniac.2026.30042

Resumo

Dyscalculia is a specific learning disorder that affects numerical processing and mathematical reasoning. It remains underrecognized in educational contexts due to limited teacher training and restricted access to specialized knowledge. This scenario contributes to the misinterpretation of students’ difficulties, delaying appropriate pedagogical interventions and exacerbating educational inequalities. This study presents DISCALCULAI, a system designed to support teachers’ literacy regarding dyscalculia and assist in identifying its indicators within educational environments. Its approach is grounded in Natural Language Processing (NLP) techniques, particularly Retrieval-Augmented Generation (RAG), to provide reliable and contextually relevant responses based on previously selected scientific literature. The system integrates text processing, semantic representation generation, vector indexing, and the use of Large Language Models (LLMs), enabling interaction through natural language queries. Unlike approaches based solely on generative models, the proposed architecture delivers evidence-based responses supported by retrieved information, reducing the occurrence of inaccurate outputs. To evaluate the system’s effectiveness, an experiment was conducted comparing the RAG-based approach with a standalone language model using a set of 31 domain-specific queries. The results indicate statistically significant improvements in semantic similarity, groundedness, and error rate (p = 0.0226), demonstrating that the integration of retrieval mechanisms substantially enhances the relevance and contextual quality of the responses. By contributing to the recognition of dyscalculia indicators, DISCALCULAI promotes more inclusive educational environments and highlights the role of technology in reducing barriers to access specialized knowledge.

Referências

Bastos, J. A. (2016). Discalculia: Transtorno Específico da Aprendizagem em Matemática. InterSaberes, Curitiba.

Bernardi, J. and Stobäus, C. (2011). Discalculia: conhecer para incluir. Revista Educação Especial, 24(39):47–60.

Bhushan, S., Arunkumar, S., Eisa, T. A. E., Nasser, M., Singh, A. K., and Kumar, P. (2024). Ai-enhanced dyscalculia screening: A survey of methods and applications for children. Diagnostics, 14(13):1441.

Campos, M. N. (2025). Parkinsonbot: chatbot para auxílio e acessibilidade sobre a doença de parkinson. Trabalho de conclusão de curso, Graduação em Ciência da Computação. Instituto de Informática, Universidade Federal do Rio Grande do Sul, Porto Alegre.

da Silva Brum, E. and Lara, I. C. M. (2019). Discalculia: um estudo sobre as dificuldades de aprendizagem em matemática. Revista de Educação, 44(2):1–15.

Dettmers, T., Pagnoni, A., Holtzman, A., and Zettlemoyer, L. (2023). Qlora: Efficient finetuning of quantized llms. In Advances in Neural Information Processing Systems, volume 36, pages 10088–10115.

Dhingra, K., Aggarwal, R., Garg, A., Pujari, J., and Yadav, D. (2022). Mathlete: An adaptive assistive technology tool for children with dyscalculia. Disability and Rehabilitation: Assistive Technology, 19(1):9–15.

Gao, Y., Xiong, Y., Gao, X., Jia, K., Pan, J., Bi, Y., Dai, Y., Sun, J., and Wang, M. (2023). Retrieval-augmented generation for large language models: A survey. arXiv preprint arXiv:2312.10997.

García, J. N. (1998). Manual de Dificuldades de Aprendizagem: Linguagem, Leitura, Escrita e Matemática. Artes Médicas, Porto Alegre.

Hu, E. J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Wang, L., and Chen, W. (2022). Lora: Low-rank adaptation of large language models. In International Conference on Learning Representations.

Johnson, J., Douze, M., and Jégou, H. (2019). Billion-scale similarity search with gpus. IEEE Transactions on Big Data, 7(3):535–547.

Jurafsky, D. and Martin, J. H. (2025). Speech and language processing. [link]. 3rd Edition Draft.

Kucian, K., McCaskey, U., von Aster, M., and Tuura, R. O. (2021). Dyscalculia: A neurocognitive perspective. Developmental Medicine & Child Neurology, 63(4):379–385.

Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., Küttler, H., Lewis, M., Yih, W., Rocktäschel, T., Riedel, S., and Kiela, D. (2020). Retrieval-augmented generation for knowledge-intensive nlp tasks. In Advances in Neural Information Processing Systems, volume 33, pages 9459–9474.

Reimers, N. and Gurevych, I. (2019). Sentence-bert: Sentence embeddings using siamese bert-networks. In Proceedings of EMNLP-IJCNLP, pages 3982–3992.

Shalev, R. (1998). Developmental dyscalculia. Journal of Child Neurology, 13(6):281–286.

Wolf, A. (2023). Tecnologias nas metodologias docentes para alunos com discalculia do desenvolvimento. Educação Matemática Sem Fronteiras, 5:76–88.