創新是聯合國永續發展目標（SDGs）極為重要的推動力，然而，全球社會如何看待創新對SDGs發展的關係，以及公眾對SDGs創新的參與度及情感仍未得到充分的了解。由於社群媒體用戶參與度、關注度和情感態度是不同的面向，因此，為了從各種角度了解永續發展目標的創新背景，一個混合的社群媒體探勘方法被應用於Twitter用戶的生成內容，以建立對三個主要目的更廣泛的理解: （1）在17個SDG的每個維度中社群媒體用戶對相關創新的參與度；（2）識別社群媒體用戶關注的主題；（3）在所探究的關注主題下衡量社交媒體用戶的情感態度。本研究亦在不同年份全面追?社群媒體用戶對某一主題的參與度、關注度和情感態度的變化。資料分析使用Python Pandas的編碼庫，將檢索到涉及SGDs內的創新的推文分配到17個SDGs；因此，17個SDGs組的最愛推特貼文和具傳播力的推特貼文被定義為參與度的分析。此外，象限分佈的參與度分析檢驗了社群媒體用戶在不同年份對17個SDGs中的每一個創新的參與度和反應。基於Hashtag標籤的網絡分析則提出了SDGs中創新的基本主題。而潛在語義分析是一種主題建模方法，它從社交媒體用戶的推文中提取有關SDGs創新的主題；據此，使用基於Python的Vader庫發現這些被發掘的關注話題下的情感態度。本研究揭示了社群意見、參與度及情感之間關聯的複雜性，這在以前的研究中是未被充分開發的；同時也讓人們注意到解釋的潛在因素，即不同類型的參與度分析可能會突出社交媒體用戶如何參與和對特定主題做出反應的不同角度。Hashtag標籤的結構化網絡鏈接也被發現證實了推文的主題分析結果。本研究亦建議，主題分析應逐年地進行參與度分析和情感分析，從而能夠深入探索SDGs創新的現狀。;Innovation is the imperative driving force of the Sustainable Development Goals (SDGs). However, inquiries into how global societies perceive the crucial role of innovation upon the progress of SDGs, and how the social public’s engagement and sentiment is expressed toward different dimensions of innovation within SDGs remain under-tapped. Since the engagement, the concerns, and the sentiment of social media users are completely different concepts. Therefore, in order to understand the SDGs innovation context from various perspectives, a hybrid social media mining approach is implemented on knowledge extracted from Twitter’s user-generated content to establish broader understanding on three major purposes: (1) disclosing the social media users’ engagement on innovation within each individual dimension of 17 SDGs; (2) identifying topics of social media users’ concerns, thereby (3) measuring the social media users’ sentiment beneath the explored topics of concern. This approach comprehensively tracks how social media users’ engagement, concerns, and sentiment upon a certain topic change in different year. Retrieved tweets addressing innovation within SDGs are assigned to 17 SDGs groups using a codebook implemented by Python Pandas; thereby, favorite tweets and viral tweets of the 17 SDGs groups are defined for the preliminary descriptive analysis of engagement. The subsequent engagement analysis using quadrant distribution examines the participation and reaction of social media users toward the innovation within each of the 17 SDGs in different years. The fundamental topical themes of innovation within SDGs are presented by the hashtags-based network analysis. Latent semantic analysis, a topic-modeling approach, extracts topics of concern from social media users’ tweets regarding SDGs innovation; whereby the sentiment underneath those explored topics of concern is discovered using Python-based Vader library. This research unveils the complexity in association between opinion patterns, engagement, and sentiment that is unprecedentedly addressed in previous research; and also brings attention to the potential pitfall of interpretation where different types of engagement analysis may highlight different angles of how social media users attend and react to a certain subject. The structured network linkage of hashtags is also found to corroborate the findings of topic analysis of tweets. The study suggests that engagement analysis and sentiment analysis should be conducted back-to-back wherever topic analysis is applied, thus enabling the in-depth exploration of the status quo of the investigated domain such as innovation within SDGs.
