An approach to unveiling personality trait classification from Bangla speeches of political leaders

Abstract

The power of speech lies in its ability to communicate with people. Political speeches can influence people’s minds through their persuasive language. The general public needs to know about the political personnel and their actions. They have huge responsibilities surrounding different sectors such as the education sector, healthcare sector, etc., which affect people’s daily lives, and their future might depend on it too. Speech is a way for political leaders to express themselves, and through their speeches, we will be trying to decipher their personality traits. To comprehend the personality traits through the text samples, we use the NEO-FFI personality traits (Openness, Conscientiousness, Extroversion, Agreeableness, and Neuroticism) to determine the personality of a particular political person. In our research, we mainly focus on Bangla speeches, and the data is collected from various authentic platforms in the audio format, which is then converted to a transcript. Politicians effectively utilizing their speech can influence the opinions of the general public. So, political speeches are a great way to understand the leaders’ personalities by the general masses to be more conscious and vote wisely. A comparative analysis is shown between different statistical and transformer based classifiers that are used to show the performance after training the model on our primary dataset.