Determining the orientation in choosing furniture based on social media based on data mining algorithms: Twitter example

Abstract

In parallel with the increase in internet usage, people from different parts of the world can easily convey their thoughts and feelings on social issues through social media. Millions of messages are written and read every day on various topics on a global scale through Twitter, which has an important place in these social media. While it is important to understand consumer behaviors in order to increase the competitiveness of firms, big data sources such as Twitter have multi-faceted the methods of analyzing behaviors. At the same time, developed countries allocate significant resources to data mining projects in order to have power. The use of Twitter and data mining as an alternative data source to identify trends in furniture choice has been proposed. The popular tweets with furniture using the Rapidminer and natural language processing software were gathered for ten months between May 2018 and February 2019, and natural language processing software enabled us to determine the mood of the tweets (positive and negative). Morphological analysis of the keywords in positive and negative tweets was then performed. Finally, meaningful information was obtained by utilizing the decision tree and association algorithms used in data mining. According to the decision tree algorithm, the most dominant words in the formation of positive or negative emotions were the challenge, campaign, discover and idea. As a result of the syntax of association, the most positive emotions were made with the order of words that awaken the emotions, and the opportunity was found as wood. In the same algorithm, the words that awaken the most negative emotions were listed as gloom, seedy, uncomfortable and fabric.