International Journal of Electrical and Computer Engineering (IJECE) Vol. 8, No. 5, October 2018, pp. 3523~3535 ISSN: 2088-8708, DOI: 10.11591/ijece.v8i5.pp3523-3535  3523 Journal homepage: http://iaescore.com/journals/index.php/IJECE Predicting the Brand Popularity from the Brand Metadata Bhargavi K 1 , Sathish Babu B 2 , S. S. Iyengar 3 1 Department of Computer Science and Engineering, Siddaganga Institute of Technology, Tumkur, Karnataka, India 2 Department of Computer Science and Engineering, R V College of Engineering, Bengaluru, Karnataka, India 3 Florida International University, Miami, Florida, USA Article Info ABSTRACT Article history: Received May 21, 2018 Revised Aug 20, 2018 Accepted Aug 28, 2018 Social networks have become one of the primary sources of big data, where a variety of posts related to brands are liked, shared, and commented, which are collectively called as brand metadata. Due to the increased boom in E/M- commerce, buyers often refer the brand metadata as a valuable source of information to make their purchasing decision. From the literature study, we found that there are not many works on predicting the popularity of the brand based on the combination of brand metadata and comment’s thoughtfulness analysis. This paper proposes a novel framework to classify the comment’s as thoughtful favored or disfavored comment’s, and later combines them with the brand metadata to forecast the popularity of the brand in near future. The performance of the proposed framework is compared with some of the recent works w.r.t. thoughtful comment’s identification accuracy, execution time, prediction accuracy and prediction time, the results obtained are found to be very encouraging. Keyword: Brand metadata Brand popularity MapReduce Social networks Thoughtful comment Copyright © 2018 Institute of Advanced Engineering and Science. All rights reserved. Corresponding Author: Bhargavi K, Faculty of Computer Science and Engineering, Siddaganga Institute of Technology, Tumkur, India. Phone: +91-9886280931 Email: bhargavi.tumkur@gmail.com 1. INTRODUCTION Social networks have connected billions of people all over the world, who generate big data in the form of text, image, and audio/video. This data would serve as a valuable source of information for many big data researchers [1]-[3]. Due to increased penetration of socialization in daily lives, the social networking has turned out to be a prominent platform for brand advertisements. This advertisement could be personalized based on the customer profile, demography specific interests, customer feedback, and other parameters. It has been analyzed that advertisers will spend over 50 billion dollars on social media advertising by 2020 [4]-[6]. Since subscribers have the freedom to express their opinions on social network sites, the platform can be misused to post meaningless or not-thoughtful comments over the brands [7]-[10]. In this work, a framework is proposed to predict the brand popularity based on brand metadata and comment’s analysis. This framework identifies thoughtful comments from the brand comment corpus and uses the comments to evaluate the current popularity of the brand. Then perform predictive analytics on the number of likes, the number of shares, and the number of identified thoughtful comments to predict the brand popularity status in the near future. Overall the brand popularity prediction aims to answer the question, i.e., "What popularity level the Brand B will be at future time T?". The proposed thoughtful comments identifier preprocesses the comments using Apache OpenNLP parser, and the opennlp.grok.ml.dectree class is used to identify the thoughtful comments [11]. Other metadata fields like number of likes, number of shares are combined with the number of identified thoughtful comments to forecast the popularity of the brand in near future.