Classification of user Comments on Virtual Reality Technology by Topic Modeling

Abstract

Abstract Today, with the increasing growth of the Internet and the rapid expansion of virtual space and its impressive features, including the increase in the speed of information exchange, easy and free access to information, the variety of topics, etc., people spend most of their time in cyberspace, especially in social networks are dedicated, in this regard, the opinions registered by users in virtual networks have grown day by day and become very important; Based on this, the aim of the current research is to analyze and review the opinions of Twitter users about virtual reality technology by using machine learning methods and a dictionary-based approach, which is advanced by collecting about 1 million tweets in the field of virtual reality technology by a web crawler. Data processing including the removal of stop words and links, de-wording was discussed, then Latent Dirichlet Allocation topic modeling was implemented on the data and the degree of semantic similarity between words and the distinction between topics was obtained by the coherence score and the number of topics that The one with the highest score was selected and the data were categorized into 9 topics. The Perplexity was used to evaluate the model, and its value was -9.44, which shows the effectiveness of the model. Then the situations related to virtual reality technology were named.

Keywords

• data mining
• text mining
• virtual reality technology
• topic modeling
• latent dirichlet allocation