Mastitis, an inflammation of the mammary gland in cattle, poses a significant threat to dairy industry productivity and animal welfare. Identifying the specific bacterial strains responsible for causing mastitis is crucial for effective prevention and treatment strategies. In this research, we have used a popular deep-learning technique to elucidate the underlying pathogenic strains associated with mastitis in cattle. The first step involved the extraction of k-mers from DNA sequences, followed by a one-hot encoding approach applied to represent each k-mer as a binary array, considering a predefined alphabet. SVD has been used to reduce the dimension of the features on that Continuous Bag-of-Words (CBOW) model, which was applied to the generated context pairs to learn meaningful embeddings for each k-mer. Node embeddings are obtained for each k-mer present in the model’s vocabulary; at last, Jaccard similarity was applied to the output node embeddings to ascertain the pathogenic potential of different bacterial strains in causing mastitis. The model enabled the identification of bacteria more prone to causing mastitis in cattle, and these findings could offer valuable insights into understanding the etiology of mastitis and contribute to the development of targeted interventions and therapeutics.