,
Assistant Lecturer, College of Computer Science and Information Technology, Wasit University, Kut, Wasit Governorate, Iraq
Assistant Professor, Faculty of Computer Science and Mathematics, Kufa University, Kufa, Najaf Governorate, Iraq
Citation impact prediction is complicated by the limitations of previous methods of prediction that are constrained to individual information channels such as textual features, citation networks, or bibliometrics. In this study, proposes a Multi-Channel Fusion Framework that leverages semantic, structural, and contextual information to predict citation impact. The proposed multi-channel model uses SciBERT-based text encodings obtained from titles, abstracts, and keywords, citation network-based graph embeddings for modeling relationships between publications, and features that describe author nfluence, journal quality, and source. An adaptive attention-based fusion architecture is used to learn the importance of each information channel prior to the classification of citation impact. Validate the framework using a Scopus dataset that consists of 11,543 Computer Science papers published from 2022 -2024, 145,329 reference relations, 48,957 authors, and 92,145 journal-level indicators. The experiment evaluation conducted with varying data split methods revealed that the proposed model scored an 84.33% accuracy and an 83.97% F1-score in comparison to the state-of-the-art methods for citation prediction. It is clear that the combination of textual semantics, citation graph, and bibliometric features gives a more complete representation of the research impact. The framework can help in identifying influential papers, evaluating research, and making decisions in academic settings. The future directions of research will consider multilingual data sets, citation graph development over time, explainable predictions, and interdisciplinary research.
This is an open access article distributed under the Creative Commons Attribution Non-Commercial License (CC BY-NC) License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The statements, opinions and data contained in the journal are solely those of the individual authors and contributors and not of the publisher and the editor(s). We stay neutral with regard to jurisdictional claims in published maps and institutional affiliations.
