Rulindung is a Deep-learning based online reporting system to classify Indonesia’s cyberviolence based on the text in the screenshot image. It's still a MVP.
-
Inside it is a Deep Learning model based on BERT architecture programmed in Python. Made use of pre-trained model that has trained with large Indonesian corpus (‘Indo4B’): IndoBERT, founded by IndoNLU team.
-
Fine-tuned with the compilation of Indonesian cyberbullying/hate-speech public datasets. The datasets obtained from several social media, comprise of Twitter & Instagram.
-
Implemented to Flask’s web app environment.
-
Used Tesseract OCR engine to detect text from image.
-
Demo: click here
-
Soon: research journal & make it as open-source project.