運用專利資料探討雲端運算技術發展趨勢

專利蘊含技術與商業的關鍵資訊，吾人若能運用專利資料來分析技術的發展趨勢，將對相關企業的技術發展策略之擬定上有很大的助益。過去在使用專利數據分析技術發展趨勢時，大多數是採書目計量法為主，然此方法卻有其侷限性，因此，越來越多的研究者開始採用文字探勘的方法來進行分析。而關鍵詞共現網絡(Keyword Co-occurrence Network)分析又是其中很重要的方法之一，惟該方法過往的使用上，在關鍵詞方面常採用單詞來分析，因而往往不能區分不同上下文的相同單詞之差異，且也會有多義詞和同義詞等語言差異的問題，因此，無法呈現出技術的全貌。爰此，本研究乃採用複合詞關鍵詞共現網絡分析方法，並選擇現階段非常重要的雲端運算技術領域為分析對象，分析其相關的國際專利分類號碼、公司名與專利題目關鍵詞之各年發展，以深入瞭解此領域中的企業專利佈局以及專利活動軌跡。本研究最後所得到的主要發現包括：一、2016年至2022年是專利數量的高峰期；二、此領域的專利發展約略可分為四階段，包括萌芽期（1998年至2002年）、扎根期（2003年至2009年）、發展期（2010年至2019年）和應用期（2020年至2022年）；三、雲端運算技術領域主要包含五大方面：雲端架構、大數據與人工智慧、隱私與安全、網路與通訊、應用開發與管理；四、雲端運算技術的專利關鍵詞可分為6個群組；五、近期雲端運算的專利發明大致上與機器學習、網絡安全協議、影像分析和自然語言建模等四方面有關。本研究的結果除了對於雲端運算技術趨勢的分析上提供了實務上的貢獻之外，亦在文字探勘的分析方法上提供一些學術上的貢獻。

Cloud computing is currently a prominent topic in the field of network technology. Due to the impact of the COVID-19 pandemic, the digital transformation of businesses has accelerated, driving the demand for cloud computing services. Significant business opportunities in cloud computing have intensified competition among key players. Patents contain a wealth of technical and critical business information, and analyzing patent data to understand the technology development trends in cloud computing can help companies formulate appropriate and strategic development strategies. Currently, most cloud computing research is primarily published in engineering, information systems, and technical journals, and the use of patent data to analyze technology development trends remains limited. Traditionally, the analysis of technology trends has often relied on bibliometrics, which has limitations. Therefore, in recent years, text analysis has gradually been applied to analyze patent information. To this end, keyword cooccurrence network analysis can be used. However, it is often applied to single keywords, and cannot differentiate between the same keywords used in different contexts. It also does not address issues such as homonyms and synonyms, language variations, and so forth. Therefore, this study developed a co-occurrence network analysis method that analyzes compound keywords. It was applied to international patent classification numbers, company names, and patent titles to explore the patent distribution within companies, identify patent activity tracks, and provide a reference for future corporate strategies. It was also used to establish a complete text-mining technology prediction model. There are several important findings. First, the number of patents peaked from 2016 to 2022. Second, the field of patent research can be divided into four stages: the embryonic (1998-2002), rooting (2003-2009), development (2010-2019), and application (2020-2022) periods. Third, the cloud computing field mainly encompasses five areas: cloud architecture, big data and artificial intelligence, privacy and security, network and communication, and application development and management. Fourth, the key patent keywords in cloud computing fall into six categories. Fifth, in recent years, patented inventions in cloud computing have focused on G06N 20“machine learning,”H04L 09“network security protocols,”G06T 0700“image analysis,”and G10L 1518“natural language modeling.”