文章詳目資料

復興崗學報

  • 加入收藏
  • 下載文章
篇名 正面與負面網路訊息影響國軍形象的大數據分析
卷期 113
並列篇名 How ROC Army’s Image is Affected by Positive and Negative Network Messages: A Big Data Analysis
作者 羅治傑王皜宇陳昱璟唐學明
頁次 143-180
關鍵字 國軍形象網路訊息大數據分析情緒分析關鍵詞分析文本探勘網路聲量社群活躍度正負情緒比the image of the ROC armynetwork messagesbig data analysissentiment analysiskeyword analysistext miningnetwork voice volumecommunity activitypositive and negative emotion ratio
出刊日期 201812

中文摘要

提升國軍形象一直是國防部的重點工作之一,攸關民眾對國軍的支持度,也攸關軍民關係的良窳。行動網路時代,網路訊息影響快速而強力,輿情分析有助於針對單一事件快速應變。此外,長期的輿情觀測,則有助於了解議題整體的趨勢。本研究先針對過去一年(106年)內三件影響國軍形象的事件,做網路訊息的洞察分析。正面影響軍人形象的事件有二,一是疾風快閃活動,二是形象劇「最好的選擇」;負面影響軍人形象的事件為慶富獵雷艦詐貸弊案。研究中搜集網路輿情資料,包括新聞網站、社群網站、部落格、及討論區等,進行情緒分析與文本探勘,以質化研究為主。此外,本研究針對一年來網路上與國軍形象有關的事件,加以歸納整理,探討輿情對這些事件的反應。國軍形象事件的民眾反應有二,一類是民眾對國軍功能(戰備、士氣、軍紀)的評價,二是對國軍的認同,共有四個面向,包括正面與負面的反應。經由大數據分析,我們以網路聲量、社群活躍度、及正負情緒比為量化指標,探討這些事件對國軍形象所造成的影響,以量化研究為主。分析結果顯示,正負情緒比與網路訊息的類別有顯著關連。關於士氣與軍紀方面,負面評價高於正面;關於認同方面,正面評價居多;關於戰備方面,正負面評價平衡。另外,社群活躍度高的訊息,如海軍的慶富案,負面評價高於正面;反之,社群活躍度低的訊息,如空軍基地開放,正面評價高於負面。

英文摘要

The image-improving of the ROC army has always been one of the key tasks by the Ministry of National Defense since it is closely related to the civilian support for the army and the good relationship between the army and the public. In this era of mobile device filled with all kinds of speedy and effective messages, the public opinion (sentiment) analysis helps us to respond swiftly to a particular event and to comprehend the overall trend of a concerned issue after a long-term observation. This study looks into three major events affecting the image of the ROC army in 2017 by examining their relevant messages on internet. Two of them are more positive image-affecting events including the “rapid flashing activity” and the propaganda film “Best Choice”. The other one is more negative, the Qingfu coastal minehunter loan-fraud scandal. We collect network messages from news sites, social networking sites, blogs, and discussion forums, for sentiment analysis and text mining, which are thought to be a qualitative research. In addition, other image-relevant events on internet are collected to explore the reaction from mass media. The image-related reaction from the public is measured by two dimensions: one is the evaluation of army functions which is further measured by three aspects, war preparation, morale, and military discipline; the other is the identity recognition from the public. Thus, there are totally four aspects of reactions, including positive and negative reactions. The more quantitative indicators such as network voice volume, community activity, and positive and negative emotion ratio are used in the big data analysis to explore the impact of those events on the ROC army image. Our results show that positive and negative emotion ratio is significantly related to the classification of network messages. Regarding morale and military discipline, negative evaluation is higher than positive; on the aspect of identity recognition, positive evaluation is more than negative; on war preparation, the positive and negative types of evaluations are considered equivalent. Furthermore, messages with high community activity, such as the Navy's Qingfu scandal, have more negative evaluations than positive ones; On the contrary, messages with a lower degree of social media activity, such as the opening of the air force bases, are more positively evaluated than negatively.

相關文獻