A new approach to generate a visual tweet from text message

Abstract

In this paper, we propose a new augmented communication method, called visual twitter, for the short messaging system. Particularly, in Twitter, text-based tweets in the limit of 140 characters are efficiently used in communicating with followers, but sometimes are not long enough to clearly express the author's own feeling or emotions. To deal with the author's feelings, we suggest enhancing a text tweet with an appropriate image, along with/without text. To generate an image from the text, we first analyze the text tweet. The morpheme analyzer detects the key words and then the thumbnail images related to those keywords are retrieved. The author can select appropriate images for background, avatars and objects. An intermediate image is then generated. After that, our emotion classifier determines the author's feeling in the text tweet using SVM (Support Vector Machine). Based on the emotion in the tweet, we use our own re-coloring method on the generated image. Our augmented visual communication method is implemented on the smart phone and the author can post her own visual tweet with or without text. The survey result shows that our method of generating visual tweets was favorable and users found the function enjoyable.