||In most of instant messenger services and chat rooms, the texts and the users’ emotions tend to show a positive correlation. This means that the users’ emotions predominantly affect their wording. While the users were having conversations within some applications such as business negotiations and dating services, the ability to accurately identify emotions of other party is the key to success. However, to recognize emotions from facial expressions is easier than from the textual contents. Strong mood swings may exist in a seemingly simple sentence in some context, so it is not sufficient to recognize the user’s emotion based only on the literal meaning of their words.|
In this study, I try to make use of emotion coordinates and vectors to record the trajectories of emotions. Recognizing and recording the move of emotions this way is more accurate than using basic tags as simple as “Happy”, “Angry”, “Sad” or other labels. Moreover, when calculating an emotion vector of a short message, one can take into account of the previous emotion contexts to determine the vector of the current emotion. This way, we can avoid misjudgment based only on the literal meaning. Emotion trajectories last calculated can be used to analyze the elements of emotions and compare the emotion of the users in instant messenger services.