787 The Keys, Roots, and Solutions to “SCI Supremacy” Hot! Ye Jiyuan DOI: 10.3772/j.issn.1000-0135.2020.08.001 To accurately and comprehensively understand the spirit of the recent documents issued by the Ministry of Education and the Ministry of Science and Technology on eliminating the “SCI supremacy” phenomenon, which have provoked some questions and concerns in the academic community, this study employs concept and case analysis to examine the key characteristics and foundations of SCI supremacy. Terms such as “supremacy” and “exclusivity” indicate a one-sided and absolute way of thinking, which equates evaluation of form with evaluation of content. Although they may share some characteristics, peer review focused on form is not the same as peer review focused on content. The roots of SCI supremacy include a shallow understanding of the characteristics of citations, the nature of academic evaluation, outdated thinking methods, the influence of one-sided academic rankings at home and abroad, and inability to reform universities and research management systems. Corresponding countermeasures include recognizing the quality of citations and adjusting evaluations of research accordingly; rather than simply using SCI source journals and citation data as criteria, the functions of citations and other bibliometrics should be taken into account. For more comprehensive and effective peer review, a system must be established to select, supervise, and meta-evaluate peer experts; to further improve the self-discipline, autonomy, and capability of the academic community to self-correct; and to facilitate a more relaxed and constructive academic environment. 2020 Vol. 39 (8): 787-795 [Abstract] ( 190 ) HTML (73 KB)  PDF (1157 KB)  ( 833 )

796 Opinions on New Science and Technology Evaluation Methods Hot! Chen Yunwei, Zhang Zhiqiang DOI: 10.3772/j.issn.1000-0135.2020.08.002 This paper analyzes the original intent and mission of research evaluation, international practices, the complexity of evaluation subjects, and the difficulty of eradicating and replacing evaluation systems to discuss three key issues in the practice of research evaluation: reasonable understanding of various evaluation indicators, rational use of paper indicators, and comprehensive application of other indicators. Three suggestions to improve research evaluation practices include understanding the basic principles of research evaluation, reforming the evaluation policy system, and applying indicators in a rational and limited way. 2020 Vol. 39 (8): 796-805 [Abstract] ( 126 ) HTML (91 KB)  PDF (1269 KB)  ( 1252 )

806 Combination of Scientometrics and Peer-Review in Research Evaluation: International Experiences and Inspirations Hot! Zhang Lin, Sivertsen Gunnar DOI: 10.3772/j.issn.1000-0135.2020.08.003 Scientometrics and peer review are the two most widely used methods of scientific research evaluation. From both theoretical and practical perspectives, the two methods have unique merits and drawbacks, and can complement each other if applied jointly. Focusing on the guidelines for “Encouraging comprehensive evaluation methods combining quantitative and qualitative approaches”, jointly suggested in recent research evaluation policy documents released by the Ministry of Science and Technology and the Ministry of Education of China, this paper reviews the status of international research and practical applications of both scientometrics and peer review to scientific evaluation. Further, target implementation suggestions are proposed based on international experiences and specific contexts within China. 2020 Vol. 39 (8): 806-816 [Abstract] ( 226 ) HTML (134 KB)  PDF (1337 KB)  ( 717 )

817 A Study on Chinese Terminology Recognition of Theory and Method from Information Science: Based on Deep Learning Hot! Wang Hao, Deng Sanhong, Su Xinning, Guan Qin DOI: 10.3772/j.issn.1000-0135.2020.08.004 The study of theory and method is the driving force for the continuous development of any discipline. It is important to understand the application and development of the current theories and methods in the subject area. In this paper, terminology recognition which is a branch of the task of named entities is used to study the theoretical methods of information science. About 20000 articles in the field of information science in the past 20 years are collected, and as large-scale corpus to be trained and tested in Bi-LSTM-CRFs, a model of Deep Learning. The experiments verify the model’s feasibility and explore the impact of each experimental variable on the model’s effect, in order to maximize the effect of model recognition. The results show that for complex entities such as theoretical method terms, the corpus recognition based on word segmentation is better than the word segmentation-based corpus. The length of the term also has a certain influence on the recognition effect. When the length of the term is too long (word count ≥6), the recognition effect is obviously reduced. At the same time, the training corpus quantity is positively correlated with the recognition effect. Larger corpus quantities lead to better recognition. The type and quantity of the entity directly affects the recognition result. The entity recognition with obvious word formation features is better. In the feature introduction experiment, in addition to the pinyin feature, the part of speech, the length of the word, and the feature of the word vector can improve the F1 value. The improvement of the word vector and the part of speech features are obvious. 2020 Vol. 39 (8): 817-828 [Abstract] ( 345 ) HTML (123 KB)  PDF (1913 KB)  ( 1247 )

829 Impact of Academic Social Network Platform Control on User Perceived Information Quality: An Empirical Study from ResearchGate Global Users Hot! Zhang Ning, Yuan Qinjian, Zhu Qinghua DOI: 10.3772/j.issn.1000-0135.2020.08.005 The new platform of academic social networks has been receiving extensive attention. To enhance the user??s information quality experience, platform service providers have implemented different control measures for the user-generated contents. Based on the theory of service quality evaluation, this study investigates the status of perceived information quality of academic social networks in 72 countries around the world. A total of 512 valid questionnaires empirically test the relationships between platform factors through partial least squares analysis, considering both the moderating effect of members?? social capital and the mediating effect of information quality experience, to reveal the user perception and judgment process of the platform controls. Research shows that ① platform reputation, management rules, and technical diagnostic factors can positively affect the information quality experience by varying degrees (βplatform reputation-information quality = 0.37; βmanagement rules-information quality = 0.16; βtechnical diagnostic-information quality = 0.23), and platform reputation also has a certain impact on the information quality expectations (βplatform reputation-information quality expectation=0.52). ② Individual needs also affect the information quality expectations (βindividual needs-information quality expectation = 0.12). However, the past experience of individuals does not show a direct effect on the information quality expectation in this study. ③ The influence of social capital on the relationship between information quality experience and perceived information quality has an interfering effect (βmoderating effect = -0.65); ④ Information quality experience has different degrees of partial mediation effects in the process of platform reputation and management rules, affecting users?? perceived information quality; however, no intermediate effect is observed in the technical diagnosis. 2020 Vol. 39 (8): 829-844 [Abstract] ( 247 ) HTML (190 KB)  PDF (2329 KB)  ( 885 )

845 Research on the Identification of Academic Innovation Contributions of Full Academic Texts Hot! Zhou Haichen, Zheng Dejun, Li Tianyu DOI: 10.3772/j.issn.1000-0135.2020.08.006 In line with recent efforts to dismantle the “SCI supremacy” phenomenon, automatic identification of the innovation and contribution of academic research could greatly improve and support both peer review and evaluation of representative work. This study proposes a method for recognizing academic innovation and contribution based on a combination of deep learning and rules. First, we identify the potential phrases related to innovation and scientific contribution in academic texts. Second, we build an automatic recognition model based on BERT, develop fine-grained extraction rules, and apply them to extraction of large-scale data sets. To empirically test the method, we applied it to evaluate research texts in the field of chrysanthemums, and were able to successfully identify and extract potential academic innovation and contributions. Future study could expand this method to visualize innovation and contribution by specific scholars. 2020 Vol. 39 (8): 845-851 [Abstract] ( 331 ) HTML (77 KB)  PDF (1196 KB)  ( 772 )

852 Using Full Content to Automatically Classify the Research Methods of Academic Articles Hot! Zhang Chengzhi, Li Zhuo, Chu Heting DOI: 10.3772/j.issn.1000-0135.2020.08.007 Automatic classification of the research methods used in academic papers is helpful for the evaluative analysis of these research methods in that it provides a basis for researchers to recommend or select the appropriate methods for their scholarly endeavors. Compared with using only abstracts for classification, the full content of articles contain more context regarding research methods, which is of great significance in exploring such automatic classification. This study examines the full content of 820 academic papers in the field of library and information science (LIS). Experts in the field of the LIS annotated method went through these academic papers. Subsequently, a training corpus for the classification of research methods was generated. We adopted the problem transformation method and algorithm adaptive method in the multi-label classification task. Na?ve Bayes and Support Vector Machine were used as the underlying classifiers of the problem transformation method to construct six different classification models. Meanwhile, the ML-KNN model in the algorithm adaptive method was selected to automatically classify the research methods used in the chosen articles. The experimental results showed that classification performance with the full article improved greatly when compared to using only the abstract. The Na?ve Bayes algorithm performed the best in the classifier chain strategy of the problem transformation method, and the F1 value reached 0.705. In addition, the results also demonstrated that research methods used in different academic papers are represented differently. A small training set would lead to low generalizability of automatic classification results. 2020 Vol. 39 (8): 852-862 [Abstract] ( 232 ) HTML (140 KB)  PDF (1551 KB)  ( 1286 )

863 Domain-Oriented Deep Semantic Event Generalization Hot! Cao Gaohui, Ren Weiqiang, Ding Heng DOI: 10.3772/j.issn.1000-0135.2020.08.008 Event generalization is a key step in constructing event evolutionary graphs. However, the current frameworks for clustering-based and classification-based event generalization ignore the structural characteristics and dynamic changes of domain knowledge, and are thus unsuitable for event generalization in a limited domain. In this paper, we propose a domain-oriented event generalization framework based on deep semantic matching, comprised of two modules: deep semantic computing and seed event matching. This framework can effectively solve the problems of dynamic fusion of domain knowledge and event semantic alignment. Using tourism data as an example, the event generalization framework demonstrates better accuracy, stability, and migration ability than the existing clustering- and classification-based frameworks. 2020 Vol. 39 (8): 863-871 [Abstract] ( 376 ) HTML (128 KB)  PDF (2589 KB)  ( 813 )

872 Characteristics and Evolution of the Trend of Interdisciplinarity in China: An Analysis Based on Web of Science Classifications Hot! Zeng Deming, Yu Yingjie, Wen Jinyan, Wang Yuan DOI: 10.3772/j.issn.1000-0135.2020.08.009 Interdisciplinarity has become increasingly common in the field of science. It is necessary to clarify the time-varying status of interdisciplinary studies and its growth patterns. Based on 2627604 papers published by Chinese scholars, we examine the extent of science convergence in a scientific innovation system, and its change in status over time. Moreover, using network analysis, we locate scientific combinations wherein interdisciplinarity occurs, and trace changes in the network and its strength over time. The analytical results reveal that interdisciplinarity is a prominent and dominant mega-trend at present, and nearly 50% of all new scientific innovation is a result of interdisciplinary research. In the future, interdisciplinarity may include more complex and heterogeneous combinations than today between more distant or diverse scientific fields and sectors. We found some consistencies in converged domains over time, which helps predict the future convergence of disciplines. Engineering Science and computer science have great influence on science convergence in China. This study can provide a point of reference to the government, universities, research institutions, and relevant researchers to formulate relevant policies, optimize the layout of disciplines, train interdisciplinary talents, and clarify the direction of innovation. 2020 Vol. 39 (8): 872-884 [Abstract] ( 285 ) HTML (138 KB)  PDF (10682 KB)  ( 578 )