| The retraction of problematic papers has been an effective measure for scientific communities to correct mistakes,maintain scientific research integrity,and promote scientific progress.As a special type of scientific literature,retracted papers have unique advantages in the empirical study of scientific research norms: they can reveal the anomie behaviors and errors of scientific communities in an objective and comprehensive manner.Under the severe situation of the spread of scientific anomie and frequent occurrences of retracting papers,it has been a very important research issue to analyze the characteristics,causes,and effects of retracted/retracting publications,and therefore propose practical countermeasures for constructing scientific research norms of global academic community,especially in China.There are some deficiencies in exsiting research,including incomplete research data,reasons’ classification without standardization and rationality,few research on the academic impact,and lack of the deep analysis on the problems and countermeasures of retracted/retracting publications.In view of this,the main contents of this dissertation can be summarized as follows:Firstly,a comprehensive,accurate and informative dataset of retracted journal publications worldwide from heterogeneous and multi-source data sources was constructed by computer processing technology and manual discrimination methods.On this basis,this study firstly conducted a bibliometric analysis on the documental features of the retracted publications to reveal "what the retracted publications are like",and explored whether there are specific features during identifying problematic papers by random forest model.The results of bibliometric analysis display that the number of retracted papers has been on the rise both globally and in Chinese mainland over the years."Repeat offenders" and large-scale retractions have a great influence on the trend of retraction numbers.Scientific powers have more articles withdrawn,but it does not mean that they have a higher tendency to retract.Life science is a typical high-incidence field of paper retractions,and problematic papers in interdisciplinary and application-oriented disciplines are also abundant.There is no significant relationship between paper retractions and journal impact.Hospitals are the "disaster area" of retractions in China,and present group characteristics in the production of retracted papers.For the authors,both "first-time offenders" are widespread and "repeat offenders" are concentrated,with "repeat offenders" being the focus of attention.The efficiency of journals in identifying and retracting problematic papers is improving,but the overall retraction time lag is not decreasing.The results of features importance measure display that there is basically no feature that can identify problematic papers.Secondly,with the help of content analysis,a scientific and complete classification system of reasons for retractions was established through a mixture of theoretical deduction and datadriven perspectives.On this basis,this study also analyzed the distribution of documental features to reveal "why the publications are retracted".The classification system has four levels,and the top-level are divided into four categories: misconduct,error,ethical anomie,and other categories.Scientific misconduct is the main reason for the retraction.Retractions in the USA are mainly related to research content,and the problem of fabrication and falsification is particularly serious.The retractions in Chinese mainland concentrated on publishing misconduct,but the number of image fraud has been increasing in recent years,mainly related to doctors and "paper mills".The retraction reasons of articles in international elite journals are basically related to the research content,and papers in low-impact journals are concentrated on publication misconduct.Scientific publications of "repeat offenders" are all retracted because of scientific anomie.The content of the retraction statement of "repeated offenses" related to research misconduct,especially the reason for the retraction,is obviously weakened and confused by the journal.Papers with different retraction reasons have obvious differences in retracted time-lag.The time lag in the fraudulent papers far exceeds other reasons.Next,based on the citations,this dissertation quantitatively measures the impact of retractions on the journal and academia from the perspectives of retracted publications and retracting behaviors,through a series of ways,such as designing metrics,constructing regression models,non-parametric tests,and full-text citation analysis.For the journal level,most retracted papers do not contribute much to the impact factor of their host journal,but their performance still varies at different levels of journals.Retracting behaviors also have no significant impact on the journal,but too many retractions in a single year,especially largescale,concentrated retractions,can significantly reduce the academic impact of the journal.For the academic level,there is a significant difference in the impact of papers with different retraction reasons.In general,fake papers have the highest cited frequency and greatest impact.Cited frequency of retracted papers before retraction is overall significantly higher than that of non-retracted papers.Nevertheless,retracting behaviors have a significant purifying effect on the influence of the paper.The high-impact retracted papers are highly cited,and it is related to both the value of themselves and the citing fashion in academia.Citing retracted publications does not necessarily have a negative impact or causes false cognition to the academia.Lastly,based on the results of informetric research,combined with qualitative materials such as retraction case reports and semi-structured interviews,countermeasures for scientific research norm construction are proposed from the following three aspects.In terms of the scientific knowledge production system,the dissertation advocates strengthening recording and preservation of research data,paying attention to publication of replication studies and null results,regulating citing behaviors and labeling format of references,and clearing the task division and blame allocation of authorship.As for the scientific achievement communication system,the dissertation proposes to develop and promote the application of problematic paper detection technology,optimize the mechanisms of peer review and post-publication peer review,strengthen the submission,application and enhanced publication of research data,and innovate and implement strictly retraction mechanism of problematic papers.Regarding the system of scientific research regulation and evaluation,it is recommended to establish the permanent regulatory institution managed by academic committee,implement strict procedures of investigating and determining and adopt balanced measures of punishment,regulate the industry of scientific editing and service and ban the "paper mills",carry out the evaluation system of representative works and optimize the review mechanisms of representative works,and speed up the reform of title appraisal and employment mechanisms for the clinicians in high-level hospitals. |
PDF Full Text Request