Abstract Interaction via social media involves frequent code‐mixed text, spelling errors and noisy elements, which creates a bottleneck in the performance of natural language processing applications. This proposed work is the first approach for code‐mixed Hindi‐English social media text that comprises language identification, detection and correction of non‐word (Out of Vocabulary) errors as well as real‐word errors occurring simultaneously. Each identified language (Devanagari Hindi, Roman Hindi, and English) has its own complexities and challenges. Errors are detected individually for each language and a suggestive list of the erroneous words is created. After this, a fuzzy graph between different words of the suggestive lists is generated using various semantic relations in Hindi WordNet. Word embeddings and Fuzzy graph‐based centrality measures are used to find the correct word. Several experiments are performed on different social media datasets taken from Instagram, Twitter, YouTube comments, Blogs, and WhatsApp. The experimental results demonstrate that the proposed system corrects out‐of‐vocabulary words as well as real‐word errors with a maximum recall of 0.90 and 0.67, respectively for Dev_Hindi and 0.87 and 0.66, respectively for Rom_Hindi. The proposed method is also applied for state‐of‐art sentiment analysis approaches where the F1‐score has been visibly improved.

The lifestyle led by today’s generation and its negligence towards health is highly susceptible to various diseases. Developing countries are at a higher risk of mortality due to late-stage presentation, inaccessible diagnosis, and high-cost treatment. Thermography-based technology, aided with machine learning, for screening inflammation in the human body is non-invasive and cost-wise appropriate. It requires very little equipment, especially in rural areas with limited facilities. Recently, Thermography-based monitoring has been deployed worldwide at various organizations and public gathering points as a first measure of screening COVID-19 patients. In this article, we systematically compare the state-of-the-art feature extraction approaches for analyzing thermal patterns in the human body, individually and in combination, on a platform using three publicly available Datasets of medical thermal imaging, four Feature Selection methods, and four well-known Classifiers, and analyze the results. We developed and used a two-level sampling method for training and testing the classification model. Among all the combinations considered, the classification model with Unified Feature-Sets gave the best performance for all the datasets. Also, the experimental results show that the classification accuracy improves considerably with the use of feature selection methods. We obtained the best performance with a features subset of 45, 57, and 39 features (from Unified Feature Set) with a combination of mRMR and SVM for DB-DMR-IR and DB-FOOT-IR and a combination of ReF and RF for DB-THY-IR. Also, we found that for all the feature subsets, the features obtained are relevant, non-redundant, and distinguish normal and abnormal thermal patterns with the accuracy of 94.75% on the DB-DMR-IR dataset, 93.14% on the DB-FOOT-IR dataset, and 92.06% on the DB-THY-IR dataset.