555 Relevance of Multi-Genre Information Unit: The Association between Task and Functional Unit Hot! Ma Cuichang DOI: 10.3772/j.issn.1000-0135.2021.06.001 Since the focus of information object in information organization and retrieval is changing from the whole document to the within-document information unit, it is necessary to study the correlation between the user’s need and information unit. Based on situational, cognitive, and granular attributes of genre, this paper explores the effects of information task, multi-genre functional unit and their interaction on perceived usefulness within the framework of interactive retrieval. We conducted two rounds of surveys with participants recruited by stratified sampling from the field of library and information science. The results show that perceived usefulness depends on the information task and functional unit, and their interaction effect; the frequency of online academic document usage can also affect perceived usefulness by each task. Based on the association between information task and multi-genre functional unit, this paper deepens our understanding of the situational and cognitive criteria in relevance judgment and the underlying mechanisms of task and genre relationship. Thus, this study will help improve the efficiency and effectiveness of fine-grained information retrieval and aggregation of online academic resources. 2021 Vol. 40 (6): 555-564 [Abstract] ( 109 ) HTML (142 KB)  PDF (788 KB)  ( 582 )

565 Technology Convergence and Evolution Path Detection: Technology Group Similarity Method Based on Time Series Analysis Hot! Chen Yue, Wang Kang, Song Chao, Zuo Jia, Pan Yuntao, Gao Jiping DOI: 10.3772/j.issn.1000-0135.2021.06.002 This paper presents the technology group similarity time series analysis method for technology fusion and evolution path detection to analyze technology development paths in the field of additive manufacturing. To this end, first, the study takes additive manufacturing technology patent data as the analysis object and measures the overall change trend of the technical field from the level of patent documentation, technical level, and technical domain level. Second, based on the co-occurrence principle of IPC classification number, the study uses the detection algorithm, identifies technology groups, and correlates the technology groups in adjacent time intervals through cosine similarity. Finally, visualization techniques are used to show the fusion and diffusion evolution relationships between the technology groups in different time intervals. The results of the study show that the additive manufacturing technology is undergoing a stage of rapid development, wherein the technology integration capability and inheritance is gradually enhanced. Additionally, this technical field has become relatively independent. The evolution path of the technology fusion and diffusion is clarified, primarily including additive manufacturing materials and processes, computer-aided design, and the three key paths of additive manufacturing applications. Recently, metal additive manufacturing and arc additive manufacturing have transformed into technical hot spots, and biomedical, construction, and food fields have become key technology application areas. The method presented in this study supplements the traditional IPC co-occurrence method. Furthermore, it shows the technological evolution path from a dynamic perspective, providing a new perspective and technical means for comprehensive detection of technological evolution path. 2021 Vol. 40 (6): 565-574 [Abstract] ( 203 ) HTML (92 KB)  PDF (6049 KB)  ( 676 )

575 Evaluation Model of Academic Social Website Based on CRM-BSC: Construction and Empirical Research Hot! Xiong Huixiang, Chen Qi, Dai Qinquan DOI: 10.3772/j.issn.1000-0135.2021.06.003 Based on domestic and foreign literature research and preliminary research, from the user’s perspective, this paper constructed the evaluation index system of academic social websites based on the CRM-BSC theoretical model and carried out empirical research, with a view to enrich the existing research and provide comprehensive evaluation standards and tools for academic social websites. Through literature and network research, the study designed the first- and third-level indicators of the academic social website evaluation model from the four dimensions of the CRM-BSC theoretical model form the top-to-bottom. Accordingly, the study tested and improved the evaluation indicators based on the small sample questionnaire test and used exploratory factor analysis to extract the second-level indicators from the bottom up. To this end, the formal version of the online survey questionnaire was designed, the sample data results were combined to scientifically test the evaluation indicator system, and the coefficient of variation method was used to assign the index weight. Three typical domestic and foreign academic social websites were selected for empirical research: ResearchGate, sciencenet, and muchong.com. The empirical evaluation results show that sciencenet and muchong.com continue having a certain gap compared with ResearchGate in terms of “user-centric” service provision. Overall, the current mainstream academic social websites fail to perform in the dimensions of “user value” and “user satisfaction.” Furthermore, some optimization space exists in the dimensions of “user knowledge” and “user interaction.” Finally, this paper analyzed the reasons for the shortcomings of the current academic social websites and introduced countermeasures to improve the construction of academic social websites from the four dimensions of CRM-BSC model. 2021 Vol. 40 (6): 575-589 [Abstract] ( 117 ) HTML (193 KB)  PDF (1357 KB)  ( 584 )

590 Analyzing the Time-lag Effect in Scientific Research Within the Same Field at Home and Abroad: Focusing on Data Mining Hot! Tan Chunhui, Xiong Mengyuan DOI: 10.3772/j.issn.1000-0135.2021.06.004 There is a difference in the scientific research level seen between domestic and international academic systems of the same scientific field. By analyzing the time-lag effect in topics and number of publications of domestic and international core journal articles in the same field, horizontal comparison in the field can be performed and the level of scientific research development can be measured. To begin with, the method and procedure of measuring the topic-based and number-based time lag in articles from domestic and international core journals are proposed. Then, taking the field of data mining as an example, literature records of core journal articles included in the CNKI (China National Knowledge Infrastructure) and WoS (Web of Science) databases from 1996 to 2019 are collected. First, the LDA (latent Dirichlet allocation) model is used for topic extraction from the literature by time-slicing, and similarity is applied on the extracted topics to measure the topic-based time-lag effect, which helps to reveal the most significant lagging direction and period in the field of Data Mining. Second, the ARDL (auto-regressive distributed lag) model is used to model and analyze the time series, which is composed of the data on the number of annual publications of domestic and international journals. By using this model, the most significant lagging coefficient can be found to identify the corresponding number-based lagging direction and period. Results suggest that in the field of Data Mining, domestic research lags behind international research with a topic-based time-lag of 3 years, with 38.6% of domestic topics lagging behind. Domestic journals also lag behind international journals in terms of number of publications with a time lag of 5 years, but the lag coefficient turns out to be 1.431913. The results prove that the proposed measurement methods for time-lag effect in different academic systems are adaptable under most circumstances. 2021 Vol. 40 (6): 590-602 [Abstract] ( 403 ) HTML (164 KB)  PDF (4101 KB)  ( 815 )

603 Exploring the “Swan Group” with “Associated- Sleeping-Beauty” in Qualified Papers Hot! Zhang Hui, Ye Ying DOI: 10.3772/j.issn.1000-0135.2021.06.005 In qualified papers represented by Nobel Prize-winning key publications, the “Swan Group” pattern with its “Associated-Sleeping-Beauty” phenomenon were discovered. Taking the Nobel Prizes in Physics and Economics as the data reference system, the research revealed that the “Swan Group” pattern can be applied to both Physics and Economics, accounting for 39.56% and 28.81% respectively. The “Swan Group” is more universal than “Black-White Swan” model, which is only applicable to natural sciences in Nobel Prize. The type distribution of “Swan Group” is consistent in Physics and Economics. Moreover, “Type 2” occupies the vast majority. Simultaneously, the phenomenon of “Associated-Sleeping-Beauty” is found in “Swan Group,” accounting for 4.00% and 6.78% in Physics and Economics, respectively, indicating that there is a higher proportion of “Associated-Sleeping-Beauty” in qualified papers. 2021 Vol. 40 (6): 603-609 [Abstract] ( 195 ) HTML (75 KB)  PDF (1274 KB)  ( 366 )

610 Patent Keyword Extraction Driven by Claim Features Hot! Yu Yan, Shang Mingjie, Zhao Naixuan DOI: 10.3772/j.issn.1000-0135.2021.06.006 The current patent keyword extraction is primarily based on general text keyword extraction methods without considering patent features. Thus, this paper proposes a patent keyword extraction method driven by patent claim features. This method includes selecting candidate keywords based on the longest common substring, removing redundant candidate keywords based on the information gain ratio, and integrating a specific degree to weigh candidate keywords. The results of real patent data demonstrate the effectiveness and feasibility of the proposed method. 2021 Vol. 40 (6): 610-620 [Abstract] ( 181 ) HTML (135 KB)  PDF (2290 KB)  ( 608 )

621 Automatic Labeling of Semantic Clauses in Research Articles Hot! Huang Wenbin, Wang Yueqian, Bu Yi, Che Shangkun DOI: 10.3772/j.issn.1000-0135.2021.06.007 Analyzing the semantic structure of research articles can be widely used to address multiple issues such as information extraction and retrieval. This paper describes the semantic structure of research articles by applying machine learning techniques to recognize the semantic types of discourse segments in these articles. We extracted the macro structure of research articles, including the syntactic and lexical information of each discourse segment as input features, and trained five models, namely support vector machines (SVM), conditional random fields (CRF), random forests (RF), gradient boost classifier (GBC), and stochastic gradient descent classifier (SGD). We integrated three best-performing models, that is, CRF, SVM, and GBC, to form a bagging model for classifying all discourse segments from the full text. Experimental results showed that our bagging model outperformed the baseline model on tasks of classifying discourse segments from full text and result sections with a higher accuracy and F-score. Furthermore, a topic-clustering experiment demonstrated the effectiveness of the model on topic detection, which is a common task in the field of text mining. 2021 Vol. 40 (6): 621-629 [Abstract] ( 211 ) HTML (105 KB)  PDF (784 KB)  ( 720 )

630 Research on the Regulation of Co-owned Privacy Information in Social Media Under the Conflict of Privacy Boundaries Hot! Zhu Hou DOI: 10.3772/j.issn.1000-0135.2021.06.008 With the popularization of social media applications, the phenomenon of privacy leakage is becoming increasingly serious. Because privacy can be leaked by the privacy owner, application providers, and others, governments, industries, and the providers have all attempted to take measures to relieve this phenomenon. However, because of the obscurity of privacy boundaries, some articles about social media include the privacy of several different users. Users often leak others’ privacy because of their low cognition ability. Aiming to define co-owned privacy management, this paper proposes tactics by users, providers, and coordinated user-providers and then tests these tactics through computational experiments. The results show that both user and provider tactics can promote relationships in social networks; provider tactics can enhance such relationships better than user tactics, user tactics require the participation of providers, and coordinated user-provider tactics not only perform well but are also feasible. Therefore, this paper proposes a coordinated user-provider mechanism of co-owned privacy management based on simulation and theoretical analyses, generating some inspiration for governing co-owned privacy. 2021 Vol. 40 (6): 630-639 [Abstract] ( 290 ) HTML (124 KB)  PDF (2042 KB)  ( 430 )

640 Emotional Load, User Stickiness, and Information Symbiosis Bias: Dynamic Analysis Based on Panel Data of Emergency Events from 2015 to 2020 Hot! Yang Changzheng DOI: 10.3772/j.issn.1000-0135.2021.06.009 To find the influence mechanism of information symbiosis, this paper explores the relationship between emotional load, user stickiness, and information symbiosis. The paper uses China’s emergency events panel data from 2015 to 2020, for example sina microblog, and employs VAR, panel data, and state space models to analyze the relationship between the emotional load, user stickiness, and information symbiosis. The findings of the research can be categorized into the following. First, in the process of information symbiosis, emotional load and user stickiness has a significant impact on information symbiosis, and the impact of emotional load is greater than that of user stickiness. In the process of emotional load fluctuations, the autocorrelation effect of emotional load has the greatest impact on information symbiosis, and the effects of information symbiosis and user stickiness are greater than other exogenous factors. Second, the marginal impact of user stickiness on information symbiosis is greater than that of emotional load; the marginal impact of information symbiosis on emotional load is greater than that of user stickiness; and the marginal impact of emotional load on user stickiness is greater than that of information symbiosis. Third, the contribution rate of emotional load to information symbiosis fluctuation is greater than that of user stickiness; the contribution rate of emotional load to user stickiness fluctuation is greater than that of information symbiosis; and the contribution rate of information symbiosis to emotional load fluctuation is greater than that of user stickiness. Fourth, the interaction effects between emotional load, user stickiness, and information symbiosis vary across demographic groups. To conclude, driving strategies and specific measures can be formulated. 2021 Vol. 40 (6): 640-655 [Abstract] ( 224 ) HTML (240 KB)  PDF (2905 KB)  ( 915 )

656 Sentiment Analysis of Online Users Based on Multimodal Co-attention Hot! Fan Tao, Wu Peng, Wang Hao, Ling Chen DOI: 10.3772/j.issn.1000-0135.2021.06.010 The sentiment of online users significantly influences the evolution and development of online public opinion events. Accurately recognizing the sentiment of online users has practical implications for managing online public opinion events. Extant studies about the sentiment analysis of online users are mostly based on texts, lacking in research on the combination of texts and images. In multimodal sentiment analysis, existing research typically unifies the comprehensive unimodal features and performs high-dimensional fusion. This can easily cause information redundancy and introduce the noise, ignoring the interaction and correlation between different modalities. Thus, we propose a sentiment analysis model based on multimodal co-attention to analyze the sentiment of online users, unifying word-guided and image-guided attention mechanisms. To this end, we conduct empirical experiments on multimodal datasets, such as “COVID-19.” The results show that the proposed model based on multimodal co-attention is superior to past models and can capture the interaction and relationship between different modalities. 2021 Vol. 40 (6): 656-665 [Abstract] ( 412 ) HTML (195 KB)  PDF (1670 KB)  ( 2478 )

666 The Influencing Mechanism of Grassroots-Level Government Trust on Non-Self-Disclosure Behavior: Taking Prevention of COVID-19 for Example Hot! Chi Maomao, Wang Junjing, Wang Weijun DOI: 10.3772/j.issn.1000-0135.2021.06.011 After the outbreak of the Coronavirus disease (COVID-19) pandemic in Wuhan, many grassroots-level governments conducted surveys on the situation of people returning from the Hubei epidemic area. However, many citizens remain unwilling to disclose (or hide) relevant personal information. Based on the S-O-R model and related literature on self-disclosure behavior, this paper explored the relationship among grassroots-level government trust, organism perception (including perceived benefits, perceived risks, and privacy concerns), and citizen's non-self-disclosure behavior. A total of 525 valid responses were collected from people who returned to their hometowns from Hubei. This study conducts an empirical analysis based on the structural equation modeling. The findings are as follows. First, grassroots-level government trust negatively and positively affects privacy concerns and perceived benefits, respectively. Second, privacy concerns positively affect perceived risks. Third, both perceived risks and privacy concerns positively affects non-self-disclosure behavior, while perceived benefits negatively affect non-self-disclosure behavior. This study further expands and enriches the existing literature on information disclosure behaviors. Finally, it provides specific policy recommendations for relevant grassroots-level governments to effectively promote citizens' self-disclosure behavior and improve government service satisfaction in the event of major public health emergencies. 2021 Vol. 40 (6): 666-678 [Abstract] ( 291 ) HTML (158 KB)  PDF (1144 KB)  ( 796 )