| 中文摘要 |
本研究旨在探討國內媒體報導人工智慧新聞社會性科學議題時,採用的新聞框架類型、論述模式、框架包裹以及新聞寫作元素的呈現情形。研究過程採內容分析與框架分析途徑,研究對象為國內「聯合新聞網」、「自由時報電子報」、「中時電子報」及「蘋果新聞網」等四大主流報社之網路電子報,樣本範圍自2019年2月起至2019年10月止,共299則新聞。研究發現,在新聞框架與子框架部分,臺灣人工智慧相關新聞內容主要呈現「工作」、「教育」、「娛樂」、「智能家庭/社交」、「安全」、「健康/醫療」、「財產交易採購」、「循環環保/交通安全車聯網」、「公共課責」、「經濟」、「相關技術應用」、「資料蒐集與使用」「責任歸屬」、「風險」及「科學/科技不確定性」等十五種框架類型,且「工作」、「教育」、「財產交易採購」、「循環環保/交通安全車聯網」、「公共課責」、「經濟」及「相關技術應用」等七種框架下又可細分出二至三種子框架。在框架使用比例上,「資料蒐集與使用」的框架頻率最高(26.1%),其次為「健康/醫療」(15.4%)及「循環環保/交通安全車聯網」(13.0%)。在框架論述模式部分,研究發現,在299則新聞樣本中,可以歸納出七種主要框架論述模式;在框架包裹部分,由框架論述模式可歸納出八種核心框架後,再統整出「進步」(75.7%)、「問題解決」(21.6%)及「危害」(2.7%)等三項框架包裹;最後,在新聞寫作元素分析部分,也發現人工智慧相關議題之科學寫作模式可以分為「主張—證據—證據」(claim-evidence-evidence, CEE)、「主張-證據-推理」(claim-evidence-reasoning, CER)、「主張—推理—證據」(claim-reasoning-reasoning, CRE)及「主張—推理—推理」(claim-reasoning-reasoning, CRR),雖然四種寫作模式的分布並無顯著差異,但其中相對完整的寫作模式(CER與CRE模式)所占總比仍超過五成以上。 |
| 英文摘要 |
The purpose of this study was to explore how the media in Taiwan present the types of frames, narrative patterns, frame packages, and writing elements about artificial intelligence (AI) news, which was a typically socio-scientific issues (SSI). The process of the research was conducted through content analysis and framing analysis. A total number of 299 from four Taiwan local daily cyber news were collected for analysis (2019/2-2019/10). The results indicated that there were fifteen frame categories used in Taiwan’s AI news, including“work”,“education”,”entertainment”,“smart home / social intercourse”,“safety”,“health / medical treatment”,“property transaction”,“environmental protection and transportation”,“public accountability”,“economic”,“technology application”,“data collection and application”,“responsibility”,“risk”, and“scientific / technological uncertainty”. Furthermore, some new sub-frames were inductively constructed. The“data collection and application”frame had the highest frequency in AI news (26.1%), followed by the frames of“health / medical treatment”(15.4%) and“environmental protection and transportation”(13.0%). In the part of narrative pattern and frame package, the results found that there were seven main types of narrative patterns and three frame packages in AI news, which included“progress”(75.7%),“problem solving”(21.6%), and“harm”(2.7%). Finally, there were four writing patterns found, including claim–evidence - evidence (CEE), claim - evidence - reasoning (CER), claim - reasoning - evidence (CRE), and claim - reasoning - reasoning (CRR). Although there was no significant difference in the distribution of the four writing patterns, the comparably better writing patterns (CER and CRE patterns) were more than 50% of the total. |