Towards Content-Dependent Social Media Platform Preference Analysis

Abstract

Social media is one of the major outcomes of progressive changes in the world of technology. The various social webs and mobile technologies have accelerated the rate at which information sharing is done, how relationships developed, and influences are held. Social media is increasingly being used by the people to help and shape the world's events and cultures with the ability to share pictures, ideas, events, etc. Further, it has transformed the way the authors interpret life and the way business is done. This article presents a decision system for selecting an appropriate social media platform (such as Facebook or Twitter) to post content with the objective to maximize the reachability of the post. The decision is made considering the domain or subject of the post and retrieving data associated with it from the web at regular time intervals. The retrieved data has been trained using logistics and K-NN regression to classify a particular instance of data and identify the platform which can provide the most reachability. The system also suggests keywords related to the topic of the post which has been mostly used in recent times.

A novel multimodal online news popularity prediction model based on ensemble learning

“…In the context of the dataset, we have taken the Mashable Inc. dataset of UCI repository but we have not taken that as it is we have extended both text feature dataset (Shivam, 2019) and image feature dataset (Bansal, 2019). Various language techniques and scores are used to upgrade the text feature dataset of all the articles' content (Kaur et al, 2020). Similarly, ResNet (Wu et al, 2019) is used to fetch Image features that are high in amount so PCA (Kim, 1996) is used for dimensionality reduction of ResNet provided approximately 10,000 features.…”