Redicting Information Reshare by People on Twitter<br/>

Name: Combustible organic materials determination and prediction of combustion properties
ISBN: 9783110572209

Please enable javascript in your browser.

Redicting Information Reshare by People on Twitter

529 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 55162 (04)
University: Sharif University of Technolog
Department: Physics
Advisor(s): Raeisi, Sadegh; Ghanbarnejad, Fakhteh
Abstract:
In this thesis, we attempt to construct a model that can predict whether someone will retweet a tweet. For this purpose, we construct a machine learning model and we use Twitter’s network features as our model’s input. We collect about 1300 random tweets and their retweets to make retweet cascades. By collecting or calculating users’ features in each retweet cascade, we construct our desired input data for our model. We test both random forest and neural networks as our machine learning section of the model. Random forest is the most accurate of the two models, predicting retweet actions with an accuracy of 0.89. Additionally, we find out that two features of the network have the greatest impact on predicting user retweets. The first feature is the number of followers the user who tweeted or other earlier retweeters have. The second one is the posting rate of tweets by the user who follows the person who tweeted or other earlier retweeters.This study can provide some new aspects about information diffusion on online social media platforms like Twitter.
Keywords:
Machine Learning ; Information Diffusion ; Graph Theory ; Online Social Networks ; Twitter Social Network ; Neural Network