Time Analysis for Sharing on Twitter Social Network by Using Neural Networks

With the increasing popularity of social networks in recent years, millions of users interact with other users daily. Individual users and corporate useAs a result ofrs are very active on social networks. Almost all brands have social network accounts and share advertisements and announcements about their brand on these accounts. Interaction on shares of the brands in social networks plays a significant role in increasing the brand's awareness through the social network. This study presents a personalized social network analysis by examining the information of more than 3000 tweets from a corporate social network account. Most essential metrics that helped more interaction and changed on tweets are posted time at the content of twits. As a result of our study, a time recommendation strategy for Twitter is proposed which increases user interaction. Therefore finding the most appropriate time for sharing will increases social media interactions such as sharing, likes and retweets. For this reason, sharing time estimation was made using multilayer perceptron (MLP) and group method of data handling (GMDH) neural networks methods with the dataset obtained from the tweeter. As a result of the study, it was determined that more successful results were obtained with 1.8% mean absolute percentage error (MAPE) using the GMDH method.