Sentiment Analysis Based on Social Networks Using Support Expectation-Maximization for E-Commerce Applications

In recent years, the use of social networks has increased rapidly. More people and companies are becoming dependent on Data Mining (DM) technology on social networking to promote the reform of unstructured data. It may be likely to put them in a systematic model. Social media has played an important role in promoting different products, generating reviews and highlighting customers' opinions. Reviews provided by the customer help to establish a good name for the product and also help to improve the quality of the product. Reviews also help customer select the best product among available products. For this reason, the provision of a particular product’s customer reviews facilitates the false-positive product review by providing a false negative of product reviews, to make it a possible downgrade which speaks about the project organization itself. The proposed Support Expectation-Maximization (SEM) algorithm is used to analyse the reviews based on sentimental lexical words to customers select the best products to resolve this problem. It collects the dataset from E-commerce online websites or applications such as Flip-kart or Amazon. The collected datasets are trained into the pre-processing process to analyse the stemming, Tokenization and Stop-word Elimination. Then, feature selection is done to select the best features of customer reviews sentiment lexical words weightages to reduce the dimensionality. A feature selection dataset is trained into the proposed Support Expectation-Maximization (SEM) algorithm to classify the reviews about the product, and it helps select the best product.