使用GPT-3語言模型輔助探討WhatsApp隱私條款事件對用戶使用行爲與態度之影響

Abstract

在現在資訊社會中，科技已經成爲主體，知識變成客體，人們依賴行動裝置中企業的通訊軟體進行訊息交換主要管道。而當一家企業的通訊軟體主導市場，假如沒有受到有效的監管，企業與使用者間權力平衡很容易就會被打破，讓權力傾向於企業，用戶便失去自身資訊控制權，當中一些有價值的資訊可能存在被企業利用或外泄的風險，侵犯著用戶的資訊隱私權。因此，本研究藉由WhatsApp隱私條款修訂事件作爲研究議題，從過去甚少被研究的亞洲裏，當中的香港地區進行切入，使用大數據分析取徑，以推拉繫泊理論作爲研究框架使用分析社會對於企業侵犯著其資訊隱私的態度與行爲。 本研究欲探討在全球兩大行動軟體商店Google Play和AppStore中WhatsApp、Signal、Telegram、Line、WeChat共五款軟體的香港地區用戶，對於WhatsApp隱私條款修訂事件，以半年為區間前後爲期18個月，共19,090筆評論内容，透過對應使用者行爲分析、GPT-3文本分析和詞頻分析進行研究。 最終研究結果發現，（一）透過使用者行爲分析上述WhatsApp外四款可能的替代軟體中，WhatsApp用戶僅有與Signal有轉換行爲；（二）而透過GPT-3文本分析證實WhatsApp隱私條款事件是促成轉換的主要推拉因素；（三）詞頻分析的結果顯示，人際工作網路的關係是主要牽制WhatsApp用戶轉移的繫泊因素。由以上結果可以歸納出，作爲亞洲一部分的香港社會，普遍對於企業侵犯著其資訊隱私的態度是反感，但他們最終也沒有因爲這樣的態度而作出轉換行爲，而是受到亞洲國家普遍存在的集體主義特性牽制而繼續使用WhatsApp。

In the current information society, technology has become the core, and knowledge has turned into an object. People rely on communication software from businesses on their mobile devices as the primary channel for exChanging messages. However, when a company's communication software dominates the market without effective regulation, the balance of power between the company and users can easily be disrupted, leading to a power shift in favor of the company. This poses a risk of companies exploiting valuable user information, infringing upon their information privacy rights. Therefore, this study, using the WhatsApp privacy policy revision as its research topic, delves into a relatively underexplored area, particularly the Hong Kong region within Asia, employing the Push-Pull Mooring Theory as a research framework to analyze the society's attitudes and behaviors regarding corporate infringements on information privacy by big data analysis. This research aims to investigate users in the Hong Kong region of five mobile applications: WhatsApp, Signal, Telegram, Line, and WeChat, available on the two major global mobile software stores, Google Play and App Store. The study focuses on their responses to the WhatsApp privacy policy revision over a period of 18 months, divided into six-month intervals, analyzing a total of 19,090 comments through behavior sequence analysis, GPT-3 NLP, and word frequency analysis.The results show that, (1) through behavior sequence analysis, WhatsApp users exhibited a conversion behavior primarily with Signal among the four possible alternative applications outside of WhatsApp; (2) GPT-3 text analysis confirmed that the WhatsApp privacy policy revision was the primary push-pull factor behind these conversions; (3) The results of word frequency analysis indicate that peer relationships were the main restraining factors for WhatsApp users' transitions. Based on these findings, it can be inferred that the general attitude in Hong Kong, as a part of the Asian region, leans toward disapproval of corporate infringements on their information privacy. However, users did not transition primarily due to this dissatisfaction but rather due to other constraining factors, which led them to continue using WhatsApp.