New Perspectives in Science Education

Twitter, now known as X, is a leading social networking and microblogging platform that serves as a prolific data repository, capturing conversations on various topics through its extensive collection of user tweets. This diverse dataset provides new perspectives and valuable information about science, technology, engineering, arts, and mathematics (STEAM) education across various educational levels. However, the high complexity, unstructured format, and large volume of this data often pose significant challenges for researchers seeking to extract meaningful insights using qualitative or quantitative approaches [1, 2, 3].

To address these challenges, we leverage the Generative Pre-trained Transformer 4 (GPT-4), an advanced multimodal large language model (LLM), to analyze tweet data. GPT-4's advanced natural language processing capabilities allow it to understand and interpret the nuances of human language, including slang, abbreviations, and context-specific language often found in tweets [4]. GPT-4's ability to infer meaning from limited text makes it ideal for analyzing such concise and sometimes cryptic messages [5]. Additionally, GPT-4 can perform semantic analysis, identifying themes, topics, and sentiments within tweets [6]. Finally, GPT-4 has been trained on a diverse large text corpus, which includes content from various cultures, enabling it to understand and analyze tweets from a wide range of global users, which is critical given Twitter's international user base [7].

Overall, this research provides two contributions: (i) a view of the new perspectives and topics related to STEAM education and (ii) a novel approach to education-related tweet data analysis using GPT-4. The data analysis findings provide pedagogical guidance to STEAM education researchers, faculty members, administrators, and policymakers on the latest trends and main topics related to STEAM education. The generated tweet dataset can also support linguists and computer scientists working in the areas of artificial intelligence and large language models.