APPLICATION OF NATURAL LANGUAGE PROCESSING TECHNIQUES (NLP) IN URBAN CODE

Manually checking for compliance is a resource-intensive and error-prone task. Information regulatory codes can be extracted automatically using natural language processing (NLP) techniques, making compliance checking simpler and more reliable. Although the PLN is an emerging area, there is still a relatively small amount of research applying to the Portuguese language. This study investigates a script using NLP techniques for the pre-processing - first phase of information extraction- of a Brazilian regulatory code. For this, the Python programming language and the NLTK library were used. An accuracy of 68% was achieved the performance the labeller, indicating the need for improvements in the pre-processing for the Portuguese language. The main contribution of this study is to allow a greater understanding of the difficulties and probable causes of label errors in the pre-processing phase.