生成式AI在博物館數據分析之應用與效益──以國立自然科學博物館為例

Application and Benefits of Generative AI in Museum Data Analysis: A Case Study of the National Museum of Natural Science

博物館的營運數據是評估整體狀況和運作效益的關鍵指標，為策略制定提供重要依據，引領未來發展方向和改進措施。本研究結合近年熱門之生成式AI，開發建立了一套博物館客製化數據洞察平台，旨在協助博物館人員，提升營運數據管理上的效率與準確性，改善過往傳統營運報告製作過程當中，繁瑣的資料蒐集與處理步驟，以及減少人工分析的時間成本。透過導入大型語言模型（LLM）API，平台能及時提供數據洞察和預測分析，協助管理者根據數據做出決策，包含調整營運策略、參觀動線與人流佈局，評估與開發全新營運計劃之參考。平台深入分析博物館全年、暑假及寒假等三大關鍵時段的收入、參觀人數、票種等多維度營運數據，全面掌握營運績效及其變化趨勢。並透過實地訪談，比較使用平台前後的情況，結果顯示該平台顯著改善了人力成本，提升了洞察的精確度。研究亦探討了平台部署及運用時面臨的挑戰與限制，並針對未來改進方向及值得深入研究的面向，提出具體建議。

Museum operational data are critical indicators for assessing overall conditions and operational efficacy, providing essential foundations for strategic planning, guiding future development directions, and improvement measures. This study integrates the increasingly popular generative AI to develop a customized data insight platform for museums, aimed at enhancing the efficiency and accuracy of operational data management. It addresses the cumbersome data collection and processing steps in traditional operational reporting and reduces the time cost of manual analysis. By incorporating a Large Language Model (LLM) API, the platform promptly delivers data insights and predictive analyses, aiding managers in making data-driven decisions, including adjustments to operational strategies, visitor flow, and layout planning, and serves as a reference for evaluating and developing new operational plans. The platform conducts in-depth analyses of multi-dimensional operational data during three key periods: the full year, summer, and winter holidays, encompassing revenue, visitor numbers, and ticket types, to thoroughly understand operational performance and its trends. Field interviews were conducted to compare conditions before and after platform implementation, revealing significant improvements in labor costs and precision of insights. The study also explores challenges and limitations faced during platform deployment and use, offering concrete suggestions for future improvements and aspects worthy of further research.