Sentiment Classification in English Using a Self-Organizing Map Algorithm with Only the One-Dimensional Vectors and a Ruzicka Coefficient in a Parallel System

Abstract

Sentiment classification has been used in many different fields because it has many significant contributions in everyday life, such as in political activities, commodity production, and commercial activities. We have proposed a new model for big data sentiment classification by using a combination of an unsupervised learning algorithm of a machine learning with a Ruzicka Coefficient (RC) in this work. A Self-Organizing Map Algorithm (SOM) of the machine learning is used in clustering the documents of the testing data set (TES) comprising 7,500,000 documents, which are the 3,750,000 positive and the 3,750,000 negative in English, into either the positive group or the negative group of our training data set (TRA) which is 3,000,000 sentences including the 1,500,000 positive sentences and the 1,500,000 negative sentences in English. In this study, we do not use a vector space modeling (VSM). We do not use any multi-dimensional vectors according to both the VSM and many sentiment lexicons. We use many sentiment lexicons of our basis English sentiment dictionary (bESD). We use many one-dimensional vectors based on the sentiment lexicons. We use a similarity coefficient in this study. We do not use any one-dimensional vectors based on the VSM. We have achieved 88.64% accuracy of the TES. The execution time of the proposed model in a distributed network environment-DNE is less than that in a sequential system-SS. Many commercial applications and surveys of the sentiment classification can widely use the results of the proposed model.