The data produces in the last two years has outweighed all the data existing up until then. Therefore, there is a need to organize and classify this information so that its retrieval is ideally relevant and smooth. The project in hand employs text mining and machine learning techniques to offer a solution to the problem. The project enables a user to upload a document and search for the document. A graphical user Interface is developed to enable a user to upload and type his search query. The documents are stored in a database. Naïve Bayesian classification algorithm is used to classify the uploaded documents into respective categories. A novel algorithm is developed based on tf - idf and cosine similarity and used for searching the database and retrieving documents relevant to the user’s query by considering user’s personal interests.