Optimization Accuracy of Tweets for Coronavirus Pandemic using BERT Based CNN Model

It assures illustrating the results of the survey and displaying series of data that is derived from the primary collection of data. In this context, it assures descriptive form of analysis of the collected primary data. It also assures summarising the collected data and information with descriptive form of statistics. It interprets the results without any bias regarding the primary data collected from survey due to its quantitative nature. In this paper coronavirus pandemic using hybrid bidirectional encoder representations from transformers (BERT) with convolution neural network (CNN) is present. At the startof training of CNN, filters are initialized with random values and as the model gets trained by back-propagation algorithm using the errors in estimating the actual output, the weights of the filters are modified to identify specific patterns in inputs. We show that combining CNN with BERT is better than using BERT and CNN on its own, and we emphasize the importance of utilizing pre-trained coronavirus models for downstream tasks. The hybrid model is implemented python platform and calculates accuracy, precision, recall and F1-score.