部落客憂鬱傾向分析與預測

Analysis and Prediction of Blogger's Depression Tendency

憂鬱症已列為聯合國世界衛生組織視為新世紀三大疾病，與癌症、愛滋病一起蠶食鯨吞著人民的身心健康。根據聯合國世界衛生組織估計，全球目前有二到四億人口正為憂鬱症所苦，估計在亞洲至少約有五千萬的憂鬱症患者，且人數不斷上升。2020年，憂鬱症將與心臟病，成為影響人類生活甚巨的前二大疾病。根據統計，台灣地區2007年統計結果，憂鬱症盛行率約8.9%，換言之，超過兩百萬人罹患憂鬱症。40%的憂鬱症患者會有輕生或自殺的念頭，10～15%的患者因自殺而死亡。所以有效的找出有憂鬱傾向的民眾已經是一項不容忽視的醫療衛生議題。因此本研究提出一項創新的憂鬱傾向預測技術，利用部落格網誌文章自動判別部落格作者的憂鬱傾向。隨著Web 2.0社群網路（Social Network）快速興起，使用者每天在部落格寫下工作和生活的諸多苦惱與需求，雖然已有許多部落格作者的情緒分析研究，但是目前並無相關研究開始探究部落格作者的憂鬱傾向預測，本研究參考美國精神科醫學會所發表之精神疾病診斷與統計手冊第四版修訂版（DSM-IV-TR）中對重度憂鬱症的定義及從部落格作者的網誌文章分析，定義出四個憂鬱傾向的因素，其中包含事件、負面情緒、症狀和負面想法。然後利用這四項因素協助判斷部落格作者的憂鬱程度，因此本研究嘗試探究下列兩項重要研究議題，並開發相關處理技術：（1）憂鬱症患者網誌文章的憂鬱傾向與相關因素分析，（2）部落格作者的憂鬱傾向預測模型。

According to the investigation report of the Department of Health, Executive Yuan, R.O.C. in 2007, it is estimated that about 7.3% of Taiwan's population suffer from the major depressive disorder. How to identify patients with depression tendency is one of important health issues. Thus, this project tries to develop a novel technique to automatically identify the depression tendency of bloggers using their blog posts. With the fast growth of social networks, bloggers usually write daily posts with their emotion and events happened in work, home, or life. Although there are lots of research works about emotion analysis and classification, to our knowledge, there is no work focusing on prediction of blogger's depression tendency based on emotion analysis. In this project, we try to analyze key factors affecting major depressive disorder, such as negative event, negative emotion, symptom and negative thought, and then use these four factors to assist bloggers to predict depression tendency. Therefore, we focus on the investigation of the following two research issues (1) analysis of relevant factors of depression on blog posts written by patients with the major depressive disorder, (2) development of event-emotion-driven depression tendency prediction model.