Sunbird/sunbert

Machine learning offers a good way to analyze natural language over social media . We build models based on deep learning architectures - Bidirectional Encoder Representations from Transformers (BERT) to perform two downstream tasks: .Analyze posts from social media as promotional, editorial or Organic . To identify tweets as either Covid19 related or not. Identify posts as either organic, organic or Organic and to label them as 'covid19' The code and datasets can be used as a starting point for any of these tasks, with minor modification to the model architecture . We open source the dataset and source code of our model called SunBERT so that other people can utilize these techniques to their,