Advancing Systematic Literature Review Methods Using Artificial Intelligence

• In Event: AI for Equity, Access, and Innovation in Postsecondary Writing and Learning Contexts

Abstract

Artificial Intelligence (AI) offers transformative opportunities for educational research, yet its methodological potential remains underutilized in many empirical studies. This presentation explores how AI-based techniques, specifically topic modeling, Natural Language Processing (NLP), and text mining, can serve as rigorous, scalable alternatives to traditional qualitative analysis, systematic literature reviews, and student feedback evaluation.

The first strand of this work applied topic modeling to conduct a systematic literature review focused on research related to inclusive pedagogy and teaching practices. Though topic modeling has a robust history in computer science and information retrieval, it is rarely applied in educational research as a synthesis tool. This unsupervised machine learning technique enables researchers to extract latent themes and trends from large corpora of scholarly work without prior manual coding, offering an efficient and unbiased approach to organizing complex literature bases.

The second study employed NLP to analyze open-ended student feedback from a graduate-level course. While institutions often rely on closed-ended Likert-scale surveys to gauge student satisfaction, this study used NLP to extract sentiment polarity (positive, negative, neutral) and identify frequently used emotionally resonant terms in students' written feedback. This approach uncovered deeper, more nuanced insights into how students perceived instructional effectiveness and engagement, offering a richer alternative to numeric summaries.

The third case focused on text mining as a strategy to analyze reflective writing from new instructors participating in a HyFlex teaching professional development program. Weekly reflections were analyzed to determine the prevalence of thematic codes over time, allowing for a quantitative trace of evolving pedagogical concerns and priorities. This hybrid analytic strategy bridged narrative inquiry with data-driven insight, revealing patterns that would have been difficult to detect through manual analysis alone.

Across all three studies, AI-enhanced methods demonstrated the ability to yield rigorous and replicable findings more quickly and systematically than traditional qualitative approaches. These methods also mitigate certain biases inherent in human-coded data, offering greater objectivity and transparency. Importantly, AI tools do not replace human judgment but augment it, supporting researchers in exploring complex questions at scale while retaining interpretive depth.

This paper advocates for broader adoption of AI methods in educational research, not merely for automation but for conceptual advancement. AI enables researchers to uncover patterns, perspectives, and connections that conventional methods may overlook, particularly when dealing with large, unstructured datasets. As the education field grapples with increasingly complex data and interdisciplinary questions, AI presents not just a technological toolset but a methodological paradigm shift. Educators, methodologists, and scholars must develop the literacy to engage with these tools critically and creatively, positioning AI not just as a technique, but as a catalyst for innovation in research design and interpretation.

Author

• Elnara Mammadova, Purdue University