檢察機關網路聲量與情緒分析－大數據分析

Internet Opinion and Sentiment of the Prosecutorial System: A Big-Data Analysis

顧以謙(Ku Yi-Chien) ； 劉邦揚(Liu Pang-Yang)

《刑事政策與犯罪防治研究專刊》 19期 (2018/12) Pp. 22-36

https://doi.org/10.6460/CPCP.201812_(19).02

摘要

隨著網路時代來臨，手機、網路與行動上網成為現代民眾溝通與接收資訊之主要管道。如能對社群平台之網路用戶進行特定議題之民意監測，不但具有節省人力、物力、財力、調查時間之優點，更比家戶電話民意調查貼近時代進步的脈動。本研究為首次嘗試大數據分析於檢察機關執法之應用，預期透過詞頻分析監測檢察機關執法網路聲量，再從中轉化詞彙意義有效分析民眾對檢察機關執法評價之情緒性，進而比較法院、檢察、警察等機關之網路聲量與好感度，本研究所使用之研究工具為《聲量趨勢觀測》、KEYPO大數據關鍵引擎、FANSDO粉絲行為追蹤器等，監測對象則為臺灣新聞媒體及社群平台大數據資訊，網路觀測期為2個月，監測期間為1年，亦即於利用2個月的觀測期間，監測2017年6月16日至2018年6月15日每個月檢察機關執法之網路聲量變化。本研究結論為：（一）網路聲量與情緒分析可補足民意調查不足之處；（二）網路聲量負面情緒加總比例，檢察機關為48.82%、司法機關為50.13%、警察機關為36.28%；（三）全年檢察機關網路聲量約17萬筆，網路聲量自「新聞媒體」向「討論區」擴散；（四）司法改革有關「濫訴」、「檢察官一、二審輪調」議題獲得民眾網路聲量支持；（五）檢察機關網路聲量之負面情緒易受到政治重大事件的投射；（六）情緒比分析顯示，檢察官親自帶隊為維護社會治安付出之努力獲得網路輿論支持；（七）「檢察機關」網路聲量及好感度與法院、警察機關迥然有異。本研究建議：（一）宜持續應用網路輿情分析於檢察機關網路聲量與情緒分析；（二）提振「檢察機關」之形象或可從有效媒體公關的宣傳著手；（三）有效調查民眾對檢察機關執法評價之情緒性，可提昇媒體因應策略的準確性；（四）建議持續重視基層檢察官聲音，改善濫訴狀況；（五）建議強化檢察機關正義形象，移除檢察機關與政治爭議、司法不公的概念式連結；（六）有關「民眾混淆司法機關職責，導致司法滿意度降低」之因果關係假設，建議未來研究提出更多明確證據；（七）回應資訊時代來臨，建議法務部及檢察機關溢注資源，建立一套足以回應公共議題的新型自我效能檢驗機制與模型。

關鍵字

檢察機關 ； 情緒分析 ； 爬蟲程式 ； 網路聲量 ； 大數據調查 ； 時間序列分析

並列摘要

The rise of the Internet has made mobile phones, online networks and other wireless devices the essential channels for people's daily communications and information exchange in the modern society. It is an economical way to carry out opinion polls on a specific issue by investigating the subscribers of social platforms in terms of manpower, material resources, budgets and time. Compared with making household phone calls, online opinion surveys are therefore more closely keeping up with the progress of the times. This study is the first attempt to make big-data analyses on law enforcement by Prosecutors Offices (LEPO). Word frequency analyses are also made to explore public opinion of LEPO on the internet in this regard. Next, it aims to effectively analyze the public sentiments towards LEPO based on the meanings of the words used by the said subscribers. Afterward, comparisons of online voice and favorability between the courts, LEPO and police authority are made. The research tools adopted in this study include ＂Voice Trend Observation＂, KEYPO (big data key engine), and the FANSDO (fan behavior tracker) to monitor and survey big data collected from news media and social platforms in Taiwan. The author spent two months to observe and survey the monthly variations of online voice on law enforcement by LEPO from June 16, 2017, to June 15, 2018. The conclusions of this study are as follows. First, online voice and sentiment analysis can serve as supplementary information to insufficient public opinion polling. Second, LEPO accounted for 48.82% of the sum of negative sentiments reflected from online voice, judiciary 50.13% and police authority 36.28%. Third, approximately 170,000 pieces of online voice in relation to LEPO were found during the surveyed year. Online voice rippled from ＂news media＂ to ＂forums＂. Fourth, the issues concerning judicial reform such as ＂vexatious litigation＂ and ＂rotation of prosecutors for the 1st & 2nd instances＂ gained support from the public online. Fifth, the negative sentiments expressed from online voice concerning LEPO were prone to major political events. Sixth, the analysis of sentimental ratio revealed that online public opinion showed support to prosecutors when they personally led a team and made efforts to defend social security. Seventh, online voice and favorability on ＂LEPO＂ are clearly distinct from those on courts and police authority. The suggestions made in this study are described below. First, it is urged to continuously analyze the public's online voice and sentiments in connection with LEPO based on the data collected from the Internet and public opinions. Second, the image of ＂LEPO＂ can be improved via effective media PR promotions. Third, effective public sentiment analysis on law enforcement by LEPO can help map out accurate coping strategies dealing with the media. Fourth, it is necessary to keep paying intensive attention to the voice from rank-and file prosecutors and curb vexatious litigation. Fifth, it is advised that the image of justice presenting by LEPO should be strengthened. On the other hand, efforts should be made to remove the conceptual associations between LEPO and political disputes and judicial injustice. Sixth, it is suggested that researchers conducting future studies find more explicit evidence regarding the hypothesis that ＂the general public's confusion over the job responsibilities held by the judiciary have led to decreasing satisfaction of the judicial system＂. Seventh, in response to the rise of the information era, the Ministry of Justice and its affiliated prosecutors agencies should distribute more resources to establish a new set of self-efficacy examination mechanism and model in order to timely and sufficiently deal with public issues.

並列關鍵字

Prosecutors Office ； sentiment analysis ； web crawler ； online voice ； big-data survey ； time series analysis

參考文獻

陳敦源 , & 蘇孔志 . (2017). 我國政府民意調查委外辦理的現狀與未來：一個簡要的回顧式探索 [The Future of Government's Public Opinion Polling: A Concise Retrospective Review]. 中國行政評論 , 23, 106–133. https://doi.org/10.6635/cpar.2017.23(2).04

劉嘉薇 . (2017). 網路統獨的聲量研究：大數據的分析 [Public Opinion of Unification-Independence on the Web: A Big-Data Analysis]. 政治科學論叢 , 113–165. https://doi.org/10.6166/TJPS.71(113-166)

Aggarwal, C. C., & Zhai, C. (Eds.). (2012). Mining Text Data. Boston, MA: Springer US.

Caldeira, G. A. (1991). Courts and public opinion. The American Courts, 303–334.

Easton, D. (1965). A systems analysis of political life.New York: John Wiley.