|dc.description.abstract||According to the evolution of mobile devices and internet, peoples’ communication behaviors were greatly involved with Mobile Instant Messaging (MIM) after 2007; the requirements of MIM not only from personal’s daily life, but also cover enterprises and kinds of organizations.
Nowadays, organizations and teams normally use Enterprise Instant Messaging (EIM) for real time or unofficial communication purposes. It is believed that EIM makes quick responses in team communication so task efficiency and performances could be greatly improved. From another perspective, EIM is also easier to get personal emotion involved in communications because it is more closed to human’s daily life (compared to traditional e-communications, e.g. email). This might cause the team communication to be complexed if people are not rational and negative emotions.
Negative emotions usually refer to angry, sad, dysphoria etc., it covers emotions that getting people feeling hurt. Some researches point that humans’ negative emotions will impacts performance of teams. And, people could get better and gracefully handle the negative emotions if the occurrences/existences of negative emotions were informed and notified.
Sentiment analysis is the method to extract subjective information from texts by nature language processing (NLP), text mining, or other computational linguistics. It was explored to address and process the negative emotions with EIM in the research of this dissertation. In this dissertation, we propose a sentiment analysis training model and process, which is based on NLP and deep learning techniques. The purposed model and process was designed to address EIM application contexts; it is goaled to proactively info EIM user about his/her negative emotions, and producing emotion spreading relationship diagram (ESRD) for support departments and manager. The expectation of purposed model is the risked introduced by negative emotions will be decreased by explicit pointed out the existences of negative emotion.
Negative emotions can be effectively handled so that team and company get benefits from more rational communications.
To verify results of the purposed model, it was integrated to one EIM platform used by dedicated company, then volunteers was interviewed to gather feedbacks. The result reveals that users commit he/she will more be careful in communications, having more concerns if they are aware someone affected by negative emotions. Mangers also describe that ESRD and notifications are useful; problems caused by negative emotions prior processed, propagation of negative emotions were controlled, and team impacted by negation emotions reduced.