The Internet revolution has created a new way of expressing the opinion of an individual. It has become a means by which people openly express their views on various topics. These opinions contain useful information that can be used in many areas that require constant feedback from clients. The analysis of opinion and its classification into different classes of feelings gradually emerges as a key factor in decision-making. There has been extensive research on automatic text analysis for feelings such as sentiment classifiers, affects analysis, automatic survey analysis, opinion extraction, or recommendation systems . These methods generally attempt to extract the overall feeling revealed in a sentence or document, either positive or negative, or somewhere in between. However, one disadvantage of these methods is that information can be degraded, especially in texts where loss of information can also occur. The proposed method attempts to overcome the problem of loss of textual information by using well-formed training sets. In addition, the recommendation of a product or the application of a product according to the requirements of the user has met with the proposed method.