Sparse Attention Mechanism And Ensemble Strategies On Transformers [Sam-Est] for Detection Of Offensive Comments in Tamil

The necessity to develop algorithms that can rapidly recognize and distinguish between offensive and inoffensive comments brings to light the need of identifying offensive comments in Tamil. It is possible to lessen the harmful impacts of online hate speech, such as trolling, cyberbullying, and social and political polarization, by identifying and removing poisonous comments while promoting constructive online discourse. It draws attention to the negative impacts of offensive words on individuals and organizations, including psychological trauma, rifts in society, and even physical violence. The advancement of online dialogue as a whole and the defense of democratic values require the creation of effective models for the detection of hostile remarks.