Partial citation analysis of five classes of retracted papers, and devising a new four-tier citation classification system for retracted (and other) papers

As the number of retractions of the academic literature increase, there is also a bibliometric interest in appreciating the volume of citations that retracted papers received prior to, and following, a retraction. We assessed the pre- and post-citation counts of one paper each of five classes of retracted papers using the Web of Science Core Collection (WOS-CC). The papers (and their respective DOIs) were: (C1) Data or classification error (https://doi.org/10.1371/journal.pone.0204076); (C2) fraud (https://doi.org/10.1007/s00265-011-1208-0); C3) paper mill (https://doi.org/10.1155/2017/2381482); C4) tortured phrase (https://doi.org/10.1007/s12652-020-02455-4); C5) plagiarism (https://doi.org/10.1016/j.neucom.2012.06.051). C1, C2, C3, C4 and C5 were cited (September 7, 2022) 43, 45, 35, 3, and 16 times, respectively. Separately, noting the shortcomings of the WOS-CC citation classification system for citations, we propose a new four-tier classification system for citation sentiment or use (background; supportive; contradictory; manipulative) that accommodates both the functionality and sentiment of a citation to assess how a retracted paper (or indeed, any paper), is cited. We tested this new classification system using C5. We envision that these two new systems (five classes of retracted papers; four tiers of citation behavior) can further be implemented in citation databases to improve the informative value of these resources