With internet becoming a base for exchanging ideas, customer reviews for a product grows rapidly in various user groups. For popular products the reviews found in the internet could be in thousands which makes it difficult to track and understand customer opinions. Similarly the manufacturer of a product or service provider finds it difficult to analyze the market pulse. Opinion mining is an emerging area of research which summarizes the customer reviews of a product or service and express whether the opinions are positive or negative. Various methods have been proposed as classifiers for opinion mining. Naive Bayesian, Nearest Neighbor techniques, Support vector machine and Decision tree induction has been extensively studied with various modifications. In this paper we propose to score the words in the opinion using Singular value decomposition, select attributes based on information gain and propose a novel neural network algorithm for classification. The proposed neural network algorithm RP-MLP Neural network is an extension of the popular multi layer perceptron neural network. The proposed model does not have full interconnectivity between the layers and hence a smaller number of weights are required. Inputs are processed in parallel using multiple multi layer perceptron and the results recombined in the end. We apply our method on 450 opinions obtained from IMDB movie review with 300 negative opinions and 150 positive opinions. The obtained classification accuracy in the proposed neural network method is 94.16%. The output obtained is compared with other popular methods including Naïve Bayesian and Decision tree algorithms. © EuroJournals Publishing, Inc. 2011.