NLPeace is a social networking app available both on desktop and mobile. We aim to connect people and foster a safe environment free of hate and offensive content. We are leveraging natural language processing to build a strong language model that will allow for our content moderation to be automatic. Hateful content is thus nipped at the bud.
imporved the performance of the hatespeech classification models by utilizing a pretrained BERT transformer model.
was able to attain an accuracy of 96% compared to the previous model with 92%.
I also trained new models for emotion detection, which could identify joy, fear, anger, and sadness in a tweet with an accuracy of 90%.
These models were then deployed with new API calls for each new model. addresses: #322 #306