Framework for laptop review analysis

Abstract

This research aims to develop a framework which enables users to be aware that what the reviews on the laptop mentioned. The objective of the research is to propose the framework for the analysis of laptop reviews. The framework consists of four main processes: preparing data for analysis, detecting subjective text paragraphs, identifying the aspect of each text paragraph and classifying the sentiment of each text paragraphs. The subjective text paragraphs are found by detecting subjective words in the sentences of each paragraph. Then, only the subjective paragraphs will be categorized into each category by comparing with the vocabulary in each aspect domain. Finally, the sentiment of paragraphs will be classified into positive and negative opinions by the Naïve Bayes classifier. The test results show that the accuracy of the subjective detection and the aspect identification of the text paragraph are more than 90% and the accuracy of sentiment classification is more than 77%. In summary, this framework is helpful in the development of the analysis system of reviews on a laptop to help consumers in making a decision before purchasing.