Detecting emotion from short messages on Nepal earthquake

摘要

Being a trending topic on natural language processing, emotion recognition still demands research to meet the end user satisfaction. This paper presents a foundation work to detect emotions at word-level. The texts used are all related to recent Nepal catastrophe. In this partial work, we reported the collection process for the event related corpus of short-messages produced during Nepal catastrophe and the building of lists of words related to emotions. We also determine the statistics for the emotional words in Assamese, Bengali and English messages.