Instant message summarization with Emoji unicode characterset support

Darren Ramsook, Patrick Hosein, Nicholas Hosein

Abstract: Summarization techniques have traditionally achieved good performance results when summarizing sentences and documents. However, their application to instant messaging notifications have not been thoroughly examined. This research outlines a model for summarizing instant messages, through the use of text and the Emoji Unicode Characterset, for applications where screen space is limited (e.g, in smartwatches and smart bracelets). The proposed model uses a Greedy N-gram token replacement method. This method produced high quality results and was evaluated using human participants. We found that there is a decrease in the time taken to read the summarized message when compared with the original message.