A linguistic approach of sound characterisation and polarization: first steps

The activities of the "TRIPLO: TRasporti e collegamenti Innovativi e sostenibili tra Porti e piattaforme LOgistiche" project, funded with funds from the Interregional Operational Programme Italy-France Maritime 2014-2020, have as their particular goal to increase the sustainability of commercial ports and associated logistic platforms, helping to lessen noise pollution [1][2]. In some project activities, the acoustic impact on the people exposed to noise from back port activities is evaluated in connection to how each person perceives the noise. Only technical investigations, which cannot ensure a phenomena's universality in terms of perception, can objectively describe a phenomenon in environmental surveys [3]; A sound can be viewed as both a physical reality that can be measured using objective criteria and a sound perception phenomenon that is of a subjective character and related to the subject's psycho-physical-emotional state. Because these two traits are inextricably linked, it is not enough to just look at them independently. Driven by these motivations, we created questionnaires concerning the perception of sounds, the structure and first results of which can be consulted in [4] [5] [6]. In this article, in the first part we present a methodology to identify adjectives characterising each sound via TF-IDF (term frequency - inverse document frequency) [7][8][9][10]; in the second part we analyse the positive or negative emotions described by the adjectives given for each sounds with TexBlob, a sentiment analysis classifier, and subsequently we compare the results obtained with the ones shown in [6].