Increasingly, buyers turn to the Internet for reviews of goods and services. It is quite difficult for an application to keep track of all the data that is available on the web. They also come in a wide variety of forms as well as a fast-moving nature. As a result, an efficient method for classifying and analysing online reviews in the context of large data is required. Opinion mining, or sentiment analysis, is a term used to describe the process of identifying and evaluating such aggregate online data. Using both information retrieval and computational linguistic tools, sentiment analysis is a particularly hard and promising science that deals with a source's reviews. Sentiment analysis and machine learning algorithms for sentiment categorization are discussed in this paper, as well as issues in opinion mining for massive data.