Identifying propaganda from online social networks during COVID-19 using machine learning techniques.

COVID-19, affected the entire world because of its non-availability of vaccine. Due to social distancing online social networks are massively used in pandemic times. Information is being shared enormously without knowing the authenticity of the source. Propaganda is one of the type of information that is shared deliberately for gaining political and religious influence. It is the systematic and deliberate way of shaping opinion and influencing thoughts of a person for achieving the desired intention of a propagandist. Various propagandistic messages are being shared during COVID-19 about the deadly virus. We extracted data from twitter using its application program interface (API), Annotation is being performed manually. Hybrid feature engineering is performed for choosing the most relevant features.The binary classification of tweets is being performed with the help of machine learning algorithms. Decision tree gives better results among all other algorithms. For better results feature engineering may be improved and deep learning can be used for classification task.

Machine learning based approaches for detecting COVID-19 using clinical text data.

Technology advancements have a rapid effect on every field of life, be it medical field or any other field. Artificial intelligence has shown the promising results in health care through its decision making by analysing the data. COVID-19 has affected more than 100 countries in a matter of no time. People all over the world are vulnerable to its consequences in future. It is imperative to develop a control system that will detect the coronavirus. One of the solution to control the current havoc can be the diagnosis of disease with the help of various AI tools. In this paper, we classified textual clinical reports into four classes by using classical and ensemble machine learning algorithms. Feature engineering was performed using techniques like Term frequency/inverse document frequency (TF/IDF), Bag of words (BOW) and report length. These features were supplied to traditional and ensemble machine learning classifiers. Logistic regression and Multinomial Naïve Bayes showed better results than other ML algorithms by having 96.2% testing accuracy. In future recurrent neural network can be used for better accuracy.