Weather events in the mid-18th century Mediterranean maritime trade: Automatic data extraction from Naples' sea protests through machine learning techniques

The paper aims to propose a methodology for the authomatic extraction and analysis of information contained in handwritten historical documents, namely sea protests. Within this general aim, in this paper, we present the use of Convolutional Neural Networks (CNNs) to classify historical narratives of climate events contained in sea protests. On this basis the paper: Draws a first map of shipwrecks related to extreme wheater events; Discovers first evidence of (possible) past medicanes; Draws a possible timeline and maps of past medicanes