Predicting Meme Success with Linguistic Features in a Multilayer Backpropagation Network

Abstract

T he challenge of predict ing meme success has gained at t ention from researchers, largely due t o t he increased availabilit y of social media dat a. Many models focus on st ruct ural feat ures of online social net works as predict ors of meme success. The current work t akes a different approach, predict ing meme success from linguist ic feat ures. We propose predict ive power is gained by grounding memes in t heories of working memory, emot ion, memory, and psycholinguist ics. T he linguist ic cont ent of several memes were analyzed wit h linguist ic analysis t ools. T hese feat ures were t hen t rained wit h a mult ilayer supervised backpropagat ion net work. A set of new memes was used t o t est t he generalizat ion of t he net work. Result s indicat ed t he net work was able t o generalize the linguist ic feat ures in order t o predict success at great er t han chance levels (80% accuracy). Linguist ic feat ures appear t o be enough t o predict meme t ransmission success wit hout any informat ion about social net work st ruct ure.

Publication Title

Proceedings of the 37th Annual Meeting of the Cognitive Science Society Cogsci 2015

This document is currently not available here.

Share

COinS