大數據視角下的公共政策－網路輿情分析方法之應用與發展

Public Policy from the Big Data Perspective-The Application and Development of Internet Public Opinion Analysis

郭毓倫(Kyle, Yulun Kuo)

《中國地方自治》 74卷9期 (2021/09) Pp. 3-35

https://doi.org/10.6581/lsgc.202109_74(9).0002

摘要

隨著智慧化手機的普及，行動通訊技術的普及也大幅改變了傳統民意的傳遞管道，透過行動網路上網的比率也逐年攀升。結合行動通訊及社群媒體的發展，網路輿情的重要性也與日俱增。雖然網路輿情資料在使用者匿名性及資料結構與可信度上常受到質疑，但無可否認地，透過迅速、大量、多元、即時性的資料搜集，能夠快速甚至同步對於政策議題進行定位及初步民意探詢，這些都是傳統民調方法難以達成的限制。快速地面對民意，積極回應群眾的治理原則，均是政策議題快速變遷的當代不可或缺的政治技能。隨著行動通訊及社群媒體的進化，民眾接觸網路管道的普及，資料搜集方法的精進，都對網路輿情資料的品質帶來本質上的改善。本文檢閱既有之技術方法與學術脈絡，具體指出網路輿情分析應用場域及限制，強化網路輿情分析方法的研究脈絡，期望能為傳統公共政策分析技術注入循證導向的新研究方法，本研究並且發現做為技術斷面的突破點，具有人工智慧特性的自然語言處理模型可應用在網路輿情分析之中，使傳統情緒分析進化至立場分析，更接近政策分析的需求。使公部門及研究學者在公共政策分析領域上能有多樣化的選擇，更多元全面客觀地對於公共政策過程進行評估與分析。

關鍵字

公共政策 ； 網路輿情分析 ； 大數據 ； 人工智慧 ； 自然語言處理

並列摘要

With the revolution of smart-phone, the information communication technology has also greatly changed the ways that public opinion delivered. In addition, the development of mobile communications and social media, the importance of internet public opinion (IPO) is also increasing day by day. Although IPO is often questioned by its anonymity, unstructured form and reliability, it is undeniable that through rapid, massive, diverse, and real-time IPO data collection, which could possibly to quickly, or even simultaneously, identifies the real-time focusing policy issues. However, those are all limitations that are barely impossible to confront in traditional polling methods. Dealing with public opinion as soon as possible and actively responding to the publics are the critical governance principles through the rapidly changing era. With the popularization of public access to the Internet, and the improvement of data collection methods, the quality of IPO data has been substantially improved. This article will review the existing technical methods and academic papers, specifically point out the application fields and limitations of internet public opinion analysis. On the other hand, this paper brings out an evidence-based research technique into traditional polls area. Thus, the policy maker and analyst can have diversified choices in public policy analysis. Besides, evaluation of the public opinion towards public policies can be more comprehensive and objective.

並列關鍵字

Public policy ； Internet public opinion analysis ； Big data ； Artificial intelligence ； Natural language process

參考文獻

Tan, C. M., Barnett, D. J., Stolz, A. J., & Links, J. M. (2011). 〈Radiological incident preparedness: planning at the local level〉. 《 Disaster medicine and public health preparedness》, 5(S1), S151-S158.

Tu、Shih、Hsiao ＆ Kuo, (2018). 《 Towards Evidence-Based Policy Deliberation for Environmental Justice - The Case Study of Civil Forums on Nuclear Waste in Taiwan》. Paper presented at The 2019 NATSA 2019 Annual Conference, University of Washington, Seattle.

Vehovar, V., & Manfreda, K. L. (2008). 〈Overview: online surveys〉. 《 The SAGE handbook of online research methods》, SAGE Publications, Inc.

Vining, A. R., & Weimer, D. L. (1999). 〈 Inefficiency in public organizations〉. 《 International Public Management Journal》, 2(1), 1-24.

丘昌泰、朱志宏（1995）。《政策規劃》。臺北：國立空中大學。