| 摘要: | 網際網路與智慧型手機的普及，使得行動即時通訊軟體已成為多數人的生活必備社 交工具，同時也很大改變使用者的行動裝置使用習慣。然後，隨伴而來諸多結果之一， 便是企業因應內部溝通需求，進而發展出的私有雲即時溝通協作平台。
企業即時訊息 (Enterprise Instant Messaging)，主要被組織用作業務內容易溝通的手 段，而最終目標是為了利用溝通效率的提升並進一步提高工作績效。換個角度來說，即 時訊息相較於電子郵件，與人們的生活行為更為貼近，因此也更容易將個人情緒帶入到 溝通行為之間;而當人們受負面情緖影響變得不理性時，將使溝通變得更複雜。
負面情緒泛指讓人類感到難過受傷的情緒，如生氣、傷心、煩燥等。研究指出，人 在情緒負面時，容易做出錯誤的決策且更傾向破壞群體組織的協調性;若能透過有效提 醒的方式，將有機會意識到自己正處於負面情緒狀態，進而面對自我的負面情緒與其來 源，減少情緒性行為並且有效解決負面情緒。
文本情感分析 (Sentimentanalysis) 是指用自然語言處理、文本挖掘以及計算機語言 學等方法來識別和提取原素材中的主觀信息的方法。對組織與企業而言，個人或團隊領 導人情緒對於個人績效或團隊的凝聚力分別有著不同的影響性。本篇論文提出一套企業 員工情緒管理的系統，透過自然語言處理的方式，訓練一個文字情緒分析的機器人，再 藉由整合企業即時訊息平台，使得機器人從旁客觀的提醒，並且記錄各別員工每日相關 的情緒分數以供相關支援單位參考，來達到主動關懷並產生負面情緒網路圖(ESRD)來降 低個體負面情緒對企業組織的風險，從而提高企業溝通與各方效率。
本研究最終得知透過情緒的回饋提醒，部分資深或具備管理經驗的員工表示，是有 機會影響到他們在個人行為上去做調整，避免自己因為言語上的不注意而影響他人，進 而造成負面情緒的擴散;具備管理職的員工也表示，從系統的提醒上得知團隊成員處於 長期情緒負面的狀態並透過負面情緒網路圖，將能有效的使他們主動去觀察成員問題， 優先進行問題的理解與預防，提高團隊溝通效率並避免嚴重影響團隊運作的問題發生。;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. |
