Analyzing AI Educational Curricula and Textbooks: A Text Mining Approach

By Dr Silpaja Chandrasekar, PhDReviewed by Susha Cheriyedath, M.Sc.Sep 7 2023

In a paper published in the journal Applied Sciences, researchers examined the intersection of artificial intelligence (AI) and education by using text mining techniques and content analysis to analyze AI educational curricula and textbooks. This study assessed the presence of key AI concepts, topic structures, and practical tools. The research employed various analytical methods, including Term Frequency (TF), Term Frequency-Inverse Document Frequency (TF-IDF) analysis, Latent Dirichlet Allocation (LDA) topic modeling, and content analysis. This paper concluded with insights for structuring effective AI curricula by providing valuable guidance for educators.

Background

Skills like computational thinking, AI knowledge, and digital literacy are crucial for students and professionals in this digital age. AI has gained prominence in education globally due to technological advances and the data revolution. The Association for Computing Machinery (ACM), the United Nations Educational, Scientific and Cultural Organization (UNESCO), and India are among other organizations and countries that are actively working to enhance AI curricula.

Textbooks are essential for delivering these curricula and guiding teaching strategies. The analysis of textbooks, employing techniques such as TF-IDF and LDA, provides insights into the knowledge structure of AI curricula. This structure can vary significantly based on social and cultural contexts as well as the institutions responsible for curriculum development. This analysis is crucial as AI gains importance in both K-12 and higher education worldwide.

Related work

Previous studies in AI education have emphasized the global importance of developing AI competency in higher education and K-12 curricula. They have highlighted the role of governmental and non-governmental entities in shaping AI education, which impacts areas like textbooks, teacher training, and lesson planning. These studies also examined how higher education standards have adapted to AI's evolving landscape and explored K-12 AI curricula at international and national levels. Text mining methods like TF-IDF and LDA have become essential tools for analyzing educational content and understanding word importance and topics. These methods can potentially unveil valuable insights into the relationships between textbook characteristics and curriculum documents by contributing to knowledge construction.

Proposed method

This study took a multifaceted approach by combining text mining and content analysis to scrutinize AI-related textbooks and their alignment with established AI curricula. The investigation involved several key stages alongside content analysis to evaluate the depth and breadth of educational content by using text mining methods, namely, TF-IDF and LDA analysis. The foundation of this study was rooted in the analysis of eight AI textbooks, which shed light on their relative importance within the curriculum domains. Data preprocessing played a crucial role in refining text data for rigorous analysis. This paper provided insights into the harmony between AI curricula and textbooks by offering valuable recommendations for enhancing AI curriculum design.

The analytical phase involved TF analysis, TF-IDF analysis, and LDA topic modeling to unveil the importance of keywords, computing-related terms, and topic composition within the textbooks. Additionally, a content analysis framework was crafted to gauge practical training tools within AI curricula. This comprehensive approach aimed to assess the synergy between educational materials and teaching tools, which ultimately contributed to the refinement of AI curriculum design and educational resources in the dynamic field of AI education.

Experimental Analysis

The study assessed the alignment between the AI curriculum and textbooks through various analyses. TF analysis revealed discrepancies between curriculum learning elements and textbook coverage. TF-IDF analysis highlighted specific terms related to the computing field and case studies. Further, LDA topic modeling uncovered distinct topic compositions across textbooks, and content analysis examined tool utilization in curriculum implementation. These findings suggest better integrating curriculum content, textbook keywords, and practical tools to enhance AI education and meet evolving societal demands.

Based on the findings, three key directions for AI curriculum development emerge. First, the curriculum should adopt a concept-based framework aligned with competency-oriented education. This framework should systematically organize curriculum areas and content elements while emphasizing the competencies of the study in knowledge, skills, values, and attitudes. A three-layered structure should be established for efficient curriculum management and application, containing detailed areas and content elements.

Second, the curriculum content should reflect technological advancements and societal changes in computing. It should draw from established standards like ACM/ Institute of Electrical and Electronics Engineers’s (IEEE) Computer Science Curricula CCDS 2021 and CS 2013 while considering international models like the K-12 AI Curriculum. Common elements across curricula should include computing basics, AI principles, data importance, and ethical considerations. Cross-cutting concepts should permeate the curriculum.

Third, practical training in AI education should be diverse and flexible by focusing on experiences, applications, understanding principles, and ethics. Instead of prescribing specific tools, the curriculum should promote a wide range of tool availability. Additionally, flexibility in time allocation, content differentiation, topic sequencing, and adaptability to different educational levels should be prioritized to enhance curriculum implementation.

Conclusion

In summary, this study thoroughly assesses the alignment between AI curriculum and textbooks using text mining and content analysis. The findings provide a strong foundation for future investigations into curriculum-textbook relationships in the computing field. Moreover, the proposed directions for AI curriculum development are crucial for fostering AI literacy and expertise in both education and industry, which makes this study valuable for the AI era.