435 Research on the Interaction of Collaborative Order at Different Levels of Social System and Knowledge System in OKC —An Empirical Analysis Based on a VAR Model Hot! Qiu Jiangnan, Cai Chengjie, Yang Chang, Li Yan DOI: 10.3772/j.issn.1000-0135.2021.05.001 In the Web2.0 era, users in the Online Knowledge Community (OKC) have formed social and knowledge systems individually as well as socially. The interaction of these two systems has become a concern for the OKC platform. Based on the theories of self-organization and complex adaptive systems, this paper constructs a collaborative order model of social and knowledge system at a level different from the micro-level of the two systems. In this paper, we use Wikipedia, a representative OKC, as a research platform, and apply the vector autoregression (VAR) method to construct four interactive interaction models among the different levels. The research results show that the micro-level of the social system has a significant interactive effect on the meso-level and macro-level of the knowledge system, and the micro-level of the knowledge system has a significant impact on the macro-level of the social system. This study enhances the existing research on OKC system ordering, and at the same time provides inspiration for the OKC platform construction and management process. 2021 Vol. 40 (5): 435-447 [Abstract] ( 185 ) HTML (180 KB)  PDF (2535 KB)  ( 540 )

448 Emotion Diffusion, Information Cascades, and Internet Opinion Deviation: A Dynamic Analysis Based on Emergency Events Panel Data from 2015 to 2020 Hot! Yang Changzheng DOI: 10.3772/j.issn.1000-0135.2021.05.002 To determine the influence mechanism of Internet opinion deviation, this paper explores the related ideas of Internet opinion deviation. The paper uses China's emergency events panel data from 2015 to 2020 and the vector autoregressive (VAR) model, panel data model, and state-space model to analyze the relationship between emotion diffusion, information cascade, and Internet opinion deviation. The research results are as follows. First, the impact of emotion diffusion and information cascade on the bias of public opinion is significant, and the impact of emotion diffusion is greater than that of the information cascade. Emotion diffusion and the autocorrelation effect of the information cascade have a positive impact on the information cascade, and the autocorrelation lag effect of emotion diffusion and the information cascade have a significant impact on emotion diffusion. Second, the marginal effect of emotion diffusion on public opinion bias is significantly greater than that of the information cascade, that of emotion diffusion on information cascade is greater than that of the public opinion bias, and that of information cascade on emotional diffusion is greater than that of the public opinion bias. Third, the contribution rate of emotion diffusion to the fluctuation of public opinion bias is greater than that of the information cascade, and that of emotion diffusion to the fluctuation of information cascade is greater than that of the public opinion deviation. Finally, the interaction effects of emotion diffusion, information cascade, and public opinion bias vary in different demographic groups. Conclusions of the study show that it is possible to formulate related driving strategies and specific measures. 2021 Vol. 40 (5): 448-461 [Abstract] ( 418 ) HTML (209 KB)  PDF (2839 KB)  ( 758 )

462 Technology Prediction Method Based on Data Fusion and Life Cycle: Empirical Analysis of Virus Nucleic Acid Detection Hot! Zhang Yang, Lin Yuhang, Hou Jianhua DOI: 10.3772/j.issn.1000-0135.2021.05.003 The construction of an efficient prediction model depends on the selection of the appropriate basic data. Given the characteristics of virus nucleic acid detection technology, and the shortcomings of existing research, such as considering a single data source and ignoring the impact of technology life cycle, this paper proposes an improved technology prediction method based on link prediction algorithm. According to the basic data set selected by considering the technology life cycle theory, we establish a weighted co-occurrence network of technical-subject-fields. This is achieved by integrating patents and documents as data sources. The empirical research and comparative analysis is conducted through a multi-level and multi-stage prediction method. Experimental results show that under the weighted link prediction index, compared with a single data source, the prediction effect of the fusion data is significantly improved. Finally, combined with the law of technology evolution and the theory of life cycle, this study provides a valuable reference for delimiting the time range of data base, which can further raise the prediction efficiency and accuracy. 2021 Vol. 40 (5): 462-470 [Abstract] ( 312 ) HTML (89 KB)  PDF (2176 KB)  ( 577 )

471 Research on Cross-media Correlation Analysis by Fusing Semantic Features and Distribution Features Hot! Liu Zhongbao, Zhao Wenjuan DOI: 10.3772/j.issn.1000-0135.2021.05.004 Several types of media data such as text, image, video, and audio are of multi-source heterogeneous type, which leads to the problem of semantic gaps. Current researches focus mostly on text and image, presumably because it is difficult to measure the correlation between more types of media data. Therefore, we discuss performing cross-media correlation analysis by fusing the semantic features and distribution features so as to produce consistent presentation of different types of media data. The different types of media data are first vectorized and input into the proposed model. Then, bidirectional long short-term memory (BiLSTM) is utilized to extract the context information, and the feature vectors are obtained. Finally, the correlation between different types of media data is analyzed by fusing the semantic features and distribution features, and all types of media data are represented consistently. The comparative experimental results show that the method proposed in this paper performs better than several traditional methods such as CCA (canonical correlation analysis), KCCA (kernel canonical correlation analysis), and Deep-SM (deep semantic match), which indicates that the proposed method can precisely detect the semantic correlation between different types of media data. The paper offers guidance and reference for research on cross-media correlation analysis. 2021 Vol. 40 (5): 471-478 [Abstract] ( 222 ) HTML (94 KB)  PDF (1387 KB)  ( 678 )

479 The Study of Company Screening Method Based on Automatic Taxonomy Construction Hot! Huang Wenbin, Bai Haodong DOI: 10.3772/j.issn.1000-0135.2021.05.005 In the equity trading market, investors' demand for scientific and effective discovery of the company groups engaged in specific business in the new third board market, is growing. Listed companies in this market have the characteristics of small business scope, high innovation, and strong cross-cutting. It is difficult for investors to find comparable companies with similar main businesses. This study proposes a method to obtain the hierarchical division among companies, based on automatic taxonomy construction. First, a weak supervision classification method is exploited to extract business terms from the main business text disclosed in annual reports. Second, clustering is conducted over the similarity of terms to obtain one business taxonomy; companies are mapped according to the terms appearing in their reports. As our experimental results show, the proposed method can help investors discover new business concepts from the financial market, understand the underlying connection between concepts and business models, compare companies in specific fields, and find investment targets. 2021 Vol. 40 (5): 479-488 [Abstract] ( 292 ) HTML (122 KB)  PDF (1416 KB)  ( 436 )

489 Cognitive Analysis Method for Mining Innovation Points in Academic Abstracts Hot! Wen Hao, He Qianru DOI: 10.3772/j.issn.1000-0135.2021.05.006 To resolve difficulties in knowledge mining algorithms due to the diversity and richness of academic abstracts' expression innovation points, this paper proposes a cognitive analysis method for innovation points mining in academic abstracts. The method comprises the following cognitive analyses: academic abstract innovation point report, lexical semantic distribution consistency, predicate verb semantic understanding, pragmatic function classification, and syntactic implicit. The research results show that these five kinds of analyses can form five levels of abstract mining, namely information retrieval, ontology construction, semantic mining, pragmatic classification, and object hidden levels. This cognitive analysis method processes natural language expression patterns using machine learning algorithms and improves the accuracy and coverage of the classification algorithm for abstract innovation points. Furthermore, it improves the efficiency of abstract “question, method, and result” triples mining. Such an intelligent question answering system based on the triad knowledge base improves the guiding roles of theory and method. 2021 Vol. 40 (5): 489-499 [Abstract] ( 289 ) HTML (162 KB)  PDF (1357 KB)  ( 479 )

500 Identifying Topic Evolutionary Pathways through Dynamic Semantic Network Analytics Hot! Chen Xiang, Huang Lu, Ni Xingxing, Liu Jiarun, Cao Xiaoli, Wang Changtian DOI: 10.3772/j.issn.1000-0135.2021.05.007 With the rapid development of science and technology, several disciplines have shown accelerating changes and intensified cross-fusion. In this context, an important problem researchers face is how research topics can be quickly and accurately identified, evolutionary pathways and trends can be tracked, and research frontiers can be subsequently comprehended. This paper therefore proposes a method for the identification of topic evolutionary pathways based on a dynamic network. First, the dynamic keyword network is constructed via introduction of the piecewise linear representation and the Word2Vec model. Second, a community discovery algorithm is used to identify the communities in the dynamic network, and the evolutionary relationship among topics is represented via measurements of the topic similarity between adjacent time intervals. Finally, the topic evolutionary pathway is identified. This study involves empirical analyses in information science. For validation of the methodology, the results obtained via using the piecewise linear representation method are compared with those obtained via the average time-division method and also with the effect of our method with K-means and LDA in topic identification. This study can therefore provide important decision support for researchers and strategic decision-makers to perform research activities aimed at progressing in the field of study. 2021 Vol. 40 (5): 500-512 [Abstract] ( 347 ) HTML (204 KB)  PDF (6115 KB)  ( 765 )

513 Research on the Path of Patent Innovation Technology Opportunities and Their Evaluation Hot! Feng Lijie, You Hongyu, Wang Jinfeng DOI: 10.3772/j.issn.1000-0135.2021.05.008 Accurately identifying innovation technology opportunities in massive patent information is a key measure to reduce corporate innovation risk. This study relies on the multidimensional map of technology innovation to build a path for identifying innovation technology opportunities. First, we use the Latent Dirichlet Allocation (LDA) algorithm to extract technical elements and topics from many patent documents. Second, we navigate and classify technical innovation elements combined with the multidimensional map of technology innovation, and then explore multiple potential technological innovation paths and multiple innovation rules. Third, the TextRank-IDF novelty evaluation index is proposed to sort the innovation opportunities to identify innovation paths with a higher value. Finally, the technical innovation of submersible motors is considered as an example to verify the technology opportunity discovery path's effectiveness. The results show that the technological innovation path under the massive patent information constructed in this study can provide a useful reference for companies to choose innovation schemes and continuously improve innovation efficiency scientifically. 2021 Vol. 40 (5): 513-522 [Abstract] ( 368 ) HTML (121 KB)  PDF (1897 KB)  ( 819 )

523 Research on User Knowledge Collaboration Mechanism and Visualization in the Open Innovation Community Hot! Zhang Haitao, Liu Weili, Ren Liang, Liu Yan DOI: 10.3772/j.issn.1000-0135.2021.05.009 Focusing on the mechanism of knowledge collaboration and interaction among users in the open innovation community has important research value for improving knowledge innovation and enhancing the community's attraction. First, there is a summary of the research on open innovation communities at home and abroad and a proposal for the research direction of knowledge collaborative interaction mechanism. Second, constructing the process model of user knowledge collaborative interaction and detailing user knowledge collaborative interaction, which is divided into collaborative knowledge symbiosis and the process of collaborative knowledge cooperation, the user knowledge collaborative network is generated and discussed in detail. Finally, taking “pollen club” as a case, which uses UCINET to visualize the process, a strong closeness exists among the users, and knowledge resources are relatively concentrated. In addition, core users or structural hole users play a key role in knowledge innovation. 2021 Vol. 40 (5): 523-533 [Abstract] ( 401 ) HTML (107 KB)  PDF (4019 KB)  ( 529 )

534 Discovering the Innovation User Groups by Converging Knowledge Characteristics and Collaborative Attributes Hot! Tang Hongting, Li Zhihong, Zhang Shaqing DOI: 10.3772/j.issn.1000-0135.2021.05.010 The open innovation community, as a typical application of innovation including user participation, can effectively gather user wisdom and provide new ideas for enterprise in product improvement and innovation. However, the openness of the community brings with it unorganized user-generated content, which in turn leads to a low efficiency of knowledge innovation in the community. At the same time, users, as independent individuals, are often limited by their cognitive constraints. To effectively gather and organize the users’ wisdom in an open innovation community, this research proposes to identify innovation user groups with great potential for product innovation under a specific knowledge context. To this end, we utilize a super-network model in analyzing complex systems to quantify the knowledge characteristics of users as well as their collaborative attributes. A genetic algorithm is used to achieve the optimal group solution. By conducting experimental research using real data from the MIUI Community, this research proves the feasibility and effectiveness of user identification. The findings theoretically contribute to the development of knowledge quantification and user identification. In addition, this research provides decision references for community management practices, especially for collaborative innovation. 2021 Vol. 40 (5): 534-546 [Abstract] ( 164 ) HTML (199 KB)  PDF (3281 KB)  ( 523 )

547 The Origin and Development of Information Science in China Hot! Ma Feicheng, Li Zhiyuan DOI: 10.3772/j.issn.1000-0135.2021.05.011 In this paper, we review the development process of contemporary information science in China and analyze the disciplinary structure of Sino-US information science under two paradigms of “Intelligence” and “Information.” The paper aims to provide a comprehensive understanding of the disciplinary system of information science in China, which will be a basis for relevant discussions of disciplinary development. We point out that information science in China originated from the scientific and technical information service in the last century. We argue for the promotion of the coordinated development of the “Intelligence” and “Information” paradigms, and the improvement of the discipline system, academic system, and discourse system of information science in China. We also propose the development of information science with Chinese characteristics. 2021 Vol. 40 (5): 547-554 [Abstract] ( 792 ) HTML (71 KB)  PDF (1075 KB)  ( 1183 )