Nowadays, E-commerce is very popular because of information explosion. Text mining is also important for information extraction. Users are more preferable to use the convenience system from many sources such as through web pages, email, social network and so on. This system proposed the relevant words extraction method for car recommendation system from user email. In relevant words extraction, this system proposed the Rule-based approach in Compiling Technique. Context- free grammar is the most suitable for relevant words extraction. Recommendation System (RS) is a most popular tool that helps users to recommend according to their interests. This system implements efficient recommendation system by using proposed key extraction algorithm, Content-based Filtering (CBF) method and Jaccard Coefficient that will help the users who want to buy the car by providing relevant car information.