The main focus of the topic is the process of transforming a collection of unstructured text documents into structured information based on mathematical and statistical principles. To begin, we’ll look at document models via the lens of the Bernoulli method, where the existence or absence of tokens the fundamental building elements of documents forms the foundation. Multinomial document model is the center of attention in an additional issue. It resembles the Bernoulli model in many ways, but instead of using the presence flag, it uses the frequentist approach, which considers how often the tokens appear in the text. To get latent topical structure across text sources and to fine-tune with the use of machine learning, we move onto researching unsupervised topic modeling strategies in the following challenge. Finally, using unstructured data analysis, we provide a model for predicting users’moods and actions on social media. A model that may capture user behavior and mood on social media is the Behavior Dirichlet Probability Model (BDPM).