Recent year sentiment analysis plays a vital role to understand the user emotions. As comprehension of the conclusion falls into place without any issues for a human, the genuine challenge is to prepare a machine to have the capacity to do that. The massive volume of data present in the web-based services today accomplishes continuous, persistent information about any item. This information as a precious asset because for every one of the individuals who are worried about general behavior, feelings, or perspectives. The paper presented the semantic meaning of the feature extraction method using low learning method of word embedding techniques. Word2vec technologies increase the similarities between opinion words and aspect words. The CNN (Convolution Neural Network) method classify sentiment polarity with prior knowledge of words. The classification accuracy and time complexities compared with existing machine learning methods. © 2019, Institute of Advanced Scientific Research, Inc.. All rights reserved.