Development and Implementation of an IA-Based System to Automate Textual Classification on Daily Drilling Reports

Abstract

Structured daily drilling reports (DDRs) are a rich source of information that allows better planning, more accurate risk analysis, improved key performance indicators (KPIs), and contracts. However, such information is originally stored in a free-text and unstructured format, which becomes difficult for efficient data mining. With the advances of artificial intelligence (AI)-based technologies, particularly AI language models, applying such techniques over unstructured data became key in the digital transformation phenomenon. This paper presents an approach for automatic DDR classification, incorporating new techniques of artificial intelligence. The proposed classifier is already in use in Petrobras for offshore DDR classification.