Enhancing Twitter Tweet Topic Understanding through Ensemble Learning

Abstract

Since Twitter’s introduction to social media of the hashtag as a content grouping label in 2007, the symbol and its associated usage as a classification la- bel have seen widespread adoption throughout social media and other platforms. While the content of a post can be conveniently classified using said post’s hash- tags, classifying posts that do not contain hashtags proves to be a much more challenging problem. In this paper, we propose a system for identifying a post’s hashtags using only the non-hashtag terms of the post and, by extension, address the issue of classifying the contents of posts that do not contain hashtags.