2026-09-042026-09-042026-07-16SANTOS, Jobson Iuri Pinheiro dos; PINHEIRO, Lucas Vinícius Machado. Pensamento computacional na Amazônia Ribeirinha: uma metodologia híbrida de ensino baseada em lendas locais. Orientador: Carlos dos Santos Portela; Coorientador: Marcelo Saldanha Lima Filho. 2026. 61 f. Trabalho de Conclusão de Curso (Bacharelado em Sistemas de Informação) – Faculdade de Sistemas de Informação, Campus Universitário do Tocantins/Cametá, Universidade Federal do Pará, Oeiras do Pará, 2026. Disponível em:https://bdm.ufpa.br/handle/prefix/10125. Acesso em:.https://bdm.ufpa.br/handle/prefix/10125Computational Thinking (CT) has become an essential competency in basic education, but its teaching in riverside schools in the Amazon faces challenges such as lack of technological in-frastructure and teacher training. This work proposes, develops, and evaluates a hybrid meth-odology (unplugged and plugged with Scratch) based on Amazonian legends (Curupira, Map-inguari, Caipora, and Boto) to teach the four pillars of CT: Decomposition, Pattern Recognition, Abstraction, and Algorithmic Thinking, to 5th grade "A" elementary school students at Tere-zinha Gueiros School in Oeiras do Pará. Twenty-five students participated. The research, of an applied nature with a qualitative approach and quantitative elements, followed action research principles, with two workshops, one lasting 4 hours and another lasting 2 hours and 30 minutes (due to a power outage). Participant observation, analysis of productions (maps, legend com-parison cards, Scratch projects), an analytical rubric, and an 11-question Kahoot applied as a post-test were used. An oral pre-test with 5 questions verified prior knowledge: no student knew what CT, decomposition, abstraction, or pattern recognition meant; only three (18,75%) had a vague notion of algorithmic thinking. In the unplugged workshop, the card-based activity com-pared the legends of Curupira, Mapinguari, and Caipora, identifying similarities (all are forest protectors) and differences (appearance and behavior). The average of 86.93% correct answers represents a significant leap compared to the zero knowledge from the pre-test. Standouts in-clude decomposition (97.9% correct) and algorithmic thinking (88.75%). Abstraction presented the greatest difficulty (56.3% correct). Qualitative evidence indicated high engagement, collab-oration, creativity (regional phrases, custom sounds), and debugging skills. It is concluded that the hybrid approach with local legends is effective, culturally relevant, and replicable in similar contexts, although the abstraction pillar requires additional reinforcement.Acesso AbertoPensamento ComputacionalComputação DesplugadaScratchLendas AmazônicasMetodologia HíbridaEducação RibeirinhaComputational ThinkingUnplugged ComputingAmazonian LegendsHybrid MethodologyRiverside EducationCNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::METODOLOGIA E TECNICAS DA COMPUTACAOTrabalho de Curso - Graduação - MonografiaAttribution-NonCommercial-NoDerivs 3.0 Brazil