Identifying Anomalous Citations for Objective Evaluation of Scholarly Article Impact

Area of Science:

• Bibliometrics
• Scholarly Communication
• Information Science

Background:

• Traditional citation-based methods for evaluating scholarly article impact face limitations in detecting anomalous citation patterns.
• These limitations can lead to unfairness and inaccuracy in research outcome assessments.
• Conflict of Interest (COI) in citations is often overlooked, potentially skewing impact evaluations.

Purpose of the Study:

• To develop a novel algorithm, COIRank, for more accurate scholarly article impact assessment.
• To identify and mitigate the influence of anomalous citations and conflicts of interest (COI) in citation networks.
• To enhance the fairness and accuracy of research impact evaluations.

Main Methods:

• Investigated citation relationships using factors like collaboration times, time span of collaboration, citing times, and time span of citing.
• Developed the COIRank algorithm, incorporating PageRank and HITS, to distinguish citation strength and rank articles.
• Studied both direct COI and suspected COI relationships within citation networks.

Main Results:

• Approximately 80.88% of articles in the American Physical Society (APS) dataset showed co-author contributions.
• 75.55% of these articles were cited by authors from the same affiliation, highlighting the prevalence of COI.
• Experimental results demonstrated that COIRank significantly outperforms state-of-the-art solutions in impact assessment.

Conclusions:

• Conflict of Interest (COI) and suspected COI significantly impact objective scientific paper evaluation and should not be ignored.
• The COIRank algorithm provides a more accurate and fair method for assessing scholarly article impact.
• The validity of the COIRank approach is confirmed through the probability of Recommendation Intensity.