Artificial Intelligence and Informetrics based Research Evaluation: Current Trends and Challenges

The combination of Artificial Intelligence (AI) and informetrics has generated new paradigms for research evaluation by allowing scalable, data-rich, and context-sensitive examinations of academic activities. AI-driven techniques, including machine learning, natural language processing, and graph-based analytics, are progressively augmenting traditional bibliometric approaches based on citation counts, impact factors, and co-authorship networks. These developments have extended the possibilities of traditional informetric systems and enabled semantic knowledge, citation prediction, subject evolution modelling, and content-based assessment. This study critically analyses state-of-the-art AI-assisted research assessment using full-text entity extraction, academic paper grading, author ranking, science-technology connection analysis, and research proposal evaluation. We investigated the design and implementation of these systems using recent empirical research and input from leading workshops such as BIRNDL, AII, and SDP, emphasising both effective uses (for example, Semantic Scholar, Microsoft Academic Graph) and failed or constrained approaches resulting from opacity, overfitting, or a lack of contextual grounding. Although significant developments have been made, the incorporation of AI into informetrics still presents ongoing difficulties, such as model explainability, algorithmic bias, insufficient standardised standards, and restricted multilingual and multimodal coverage. Emphasising the importance of hybrid human-AI systems, inclusive datasets, open models, and policy-aligned design frameworks, we conclude with a roadmap for future research. This study lays the foundation for responsible innovation in AI-enabled research evaluation and presents a thorough analysis of the trajectory of the field.