以臺灣社群聆聽產業之剖析 探究大數據分析的侷限及倫理問題

Abstract

大數據（big data）為近幾年來最受寵的技術之一，任何產品只要冠上「大數據」三字，便如同站在科技端的最前線，各界紛紛試圖利用這項技術來挖掘具有價值的訊息，然大數據在現今的社會裡被過度炒作，甚至成為一種迷思，外界甚少了解大數據本質，以及其實際的運作流程。其中大數據應用藍圖裡，隨著巨量資料和社群媒體熱絡發展，因而快速竄紅的社群聆聽，成為一種新興產業與研究工具，其橫跨科技與社會人文科學的特徵，讓該產業的發展過程、面臨之挑戰與侷限，成為得以反映大數據與現今社會關係的縮影。

因此，本研究採取深度訪談法，訪問9位社群聆聽的內部工作者，以透過分析台灣社群聆聽產業，探究整個大數據環境與社會、政經及倫理面向所交織的意義及影響，同時檢視社群聆聽大數據現階段之發展、面臨的困境與挑戰。

本研究結果發現，大數據在爬蟲、建模、清洗，以及分析等步驟上仍具有一定程度的誤差，研究過程也會因數據工作者的專業度、社會洞察力、個人意識形態等人為變因，而存有數據結果偏差的疑慮，且大數據並非適合所有研究命題，必須搭配其他研究方法和資料相輔相成，增加研究精準度。而大多數的數據掌握在少數的企業上，形成數據獨裁的現象，互不通聯的數據猶如數據孤島，成為阻礙大數據發展的一面高牆。另外，社會大眾對大數據分析具有不正確的遐想，並呈現資訊落差的情況，被炒作的大數據熱潮讓人們試圖以大數據量化所有具象的、抽象的物質，然並非所有的事物都可以被數據化，大數據受到科技追逐賽以及市場導向的干擾，早已扭曲反映社會的初衷，成為了影響社會走向的工具。

隨著大數據發展相繼產生之資安洩漏、非法數據交易、數據侵權等問題，在台灣未設有大數據專法的情況下，僅能以現階段的其他法規進行約束，但未臻完善的規範仍具有律法無法觸及之處，數據工作者只能遵從自律原則、恪守工作倫理，但在以商業利益為導向的數據產業中，大數據倫理綱要難以得到共識與發展，巨量數據的使用規範與倫理約束也不該僅侷限於數據使用者或相關從業人員，正確的觀念與知識應該同時落實於社會，因科技所產生之問題，必須依靠社會整體的集體意識共同努力，而非單方面的檢討與限制。Big data is one of the most inexorably trending technologies in recent years. Products become avant-garde as they are claimed to use data science. Every segment of society has scrambled for big data making it overhyped and a kind of myth. In other words, the masses know little about the nature of big data and its actual operation process. Besides, among data industries, social listening rapidly draws attention with the rise of big data and social media. It has been all the rage in academia and business industries. The multidisciplined characteristics of social listening which include science and social humanities making it a suitable microcosm to reflect the problems and challenges between our modern society and data universe.

Therefore, the in-depth interview is adopted in this research to interview 9 internal workers in social listening enterprises, trying to figure out the aspects of ethical issues, dilemmas and challenges striking against the social listening industry. Meanwhile by analyzing the interview results, the phenomenon of how the entire data environment has impacted on our society is concernedly discussed.

The study findings show that it has a certain degree of error and biases in the procedures of data crawling, data modeling, data cleaning, and data analysis. The outcome of data researches would be subject to variations like data workers’ professionalism, abilities to social insights, personal ideology and so on. Additionally, being a research method, big data doesn’t fit in all research propositions. It must also be complemented by other research methods or information to enhance research accuracy. Most of the data is held by a few companies which means data dominance and fragmented data are hindering technology from moving forward. Moreover, the masses usually have incorrect reveries about big data revealing the severe information divide. People try to quantify everything with big data, but not everything can be digitized. Affected by technology race and market-oriented interference, the intention of big data to reflect society has been distorted. Big data has become a tool that influences the society.

With the development of big data, there have been problems such as security leaks, privacy issues, illegal data transactions, and data infringement. In the absence of data protections in Taiwan, related issues can only be restricted by other existing regulations. However, the incomplete binding rules are not comprehensive. Self-discipline becomes crucial for every data worker. Nevertheless, data ethics framework is hard to implement in data industries oriented by business interests. Instead of restraining data industries unilaterally, it is ideal to educate the masses on big data and work together on problems resulting from the emerging technology.