Lexical-Syntactic Evaluation of written activities based on Genetic Algorithm and Natural Language Processing: An experiment on ENEM

The use of information and communication technologies is increasingly standing out in Education. Currently, there are several technologies that provide support for distance learning. Before this scenario, there are millions of users with an huge amount of educational data, which evaluation burdens teachers and consume their time, leaving the students waiting for the evaluation of their performance. With the use of automatic correction technics performed by computers, there is a bigger fluidity of activities and a relevant decrease of work for the teachers, leaving them free to evaluate others activities. This paper presents a lexical-Syntactic Analyzer for automatic evaluation of written Activities in Portuguese using Genetic Algorithms and Natural Language Processing – NLP. The proposed system was evaluated by experiment, about 20 written activities and it found and evaluated the mistakes previously identified by human correctors. The results demonstrated a high rate of appropriate suggestions for the errors detected.