Abstract: The exponential growth of Web 2.0 concept influences the size of Internet information storage very much. People all over the world can easily interact with each other and share their emotions and experiences. This study aims to provide a way for secured opinion polling to improve the reliability of online reviews. While considering a set of data for review, careful selection of data set is needed, since there are wide chances for fake or spam opinions. Spam opinions can be posted by any persons who purposely post unreal information or by the business firm itself to improve their productivity by cheating users. Mining these reviews has become a challenging research nowadays. Many researchers concentrated on reviewing opinions for different domains. This study proposes new three tier security architecture to poll the reviews about holiday resorts. First tier creates secured customer id. Second tier is responsible for polling reviews with authorized entry with alert messages. Third tier provides restricted view to new customer who seeks reviews and read-only option of all reviews to the firm. Category wise reviews concept is useful for the customer to check their taste and for resort people to get more profit by enhancing the features to the customer. With this outcome the variations in reliability (92-62%), confidentiality (98-22%) and agility (89-42%) of opinion mining tool, with security and without security is analyzed.