Determining the sentiment polarity of product reviews has become a landmark work in natural language processing (NLP) and data science. This is perhaps easing the way of understanding and it is also easy to get good results with very simple methods (e.g. positive – negative words counting).In this paper deep architecture based sentiment analysis product reviews are done by Convolution Neural Network (CNN) and Long Short Term Memory (LSTM). The publically available online item reviews from Amazon reviews web portal are used for analysis. Experiments with neural network methods show promising outcomes.