Intelligence Theories and Methods

633 Predictive Intelligence Research Based on Data Intelligence: A New Paradigm for Predictive Intelligence Research in a New Environment Hot! Li Yang, Sun Jianjun DOI: 10.3772/j.issn.1000-0135.2024.06.001 Predictive intelligence is the core of intelligence research. This study reviews the origin and evolution of predictive intelligence based on an analysis of predictability and unpredictability and the characteristic idea of predictive intelligence research. Employing the characteristics of the data intelligence environment, this study puts forward a new paradigm called predictive intelligence based on data intelligence, and elaborates upon it according to the features of connotation characteristics, a philosophical perspective, and value significance. The study holds that the predictive intelligence research paradigm based on data intelligence has important new features such as big data support, intelligent technology application, active intervention, engineering analysis, multi-scale description, high precision orientation, and so on. In essence, this is the epistemological change of predictive intelligence brought about by the rise of “machine experience.” The new paradigm can further promote predictive intelligence research to achieve a “forward threshold” in prediction concepts, in-depth mining in prediction content, and “real world” and even “multiple intertwined worlds” in the prediction space. Indeed, the future development of predictive intelligence based on data intelligence still must pay considerable attention to the basic guarantee, systematic balance, interpretability, ethics and risks, and human wisdom support. 2024 Vol. 43 (6): 633-643 [Abstract] ( 173 ) HTML (110 KB)  PDF (751 KB)  ( 108 )

644 “Ability”and“Potential”: Research on the Dynamic Model of Interdisciplinary Knowledge Communication from the Perspectives of Ecology and Physics Hot! Ye Guanghui, Peng Ze, Li Songye, Xia Lixin DOI: 10.3772/j.issn.1000-0135.2024.06.002 An increasing number of scholars use diverse knowledge via interdisciplinary knowledge communication to solve more comprehensive and complex academic problems against the background of blurred subject boundaries and broken disciplinary barriers. In order to identify the dynamics and development direction of interdisciplinary knowledge communication behavior, this study constructs a power model of interdisciplinary knowledge communication from the perspectives of ecology and physics. First, the concepts of potential energy in physics and ecological niche in ecology are transplanted into the interdisciplinary knowledge communication field. Second, based on the researchers’ knowledge base, the interdisciplinary knowledge communication potential energy is calculated by integrating the network structure characteristics and knowledge niche width of the knowledge subject. Then, based on the potential energy difference of interdisciplinary knowledge communication and overlap degree of knowledge niche among scholars, the trend of interdisciplinary knowledge exchange among scholars is quantified, and a power model of interdisciplinary knowledge exchange from the perspective of scholars is constructed. Taking the knowledge communication network as a research object, this study describes the activity of interdisciplinary knowledge communication behaviors and the evolution process of the entire academic ecosystem and constructs a power model of the network evolution. Finally, the power model is verified by using the interdisciplinary knowledge communication network of scholars in the field of health informatics as experimental data, and the evolution process and characteristics of interdisciplinary knowledge communication in network are analyzed. 2024 Vol. 43 (6): 644-657 [Abstract] ( 170 ) HTML (167 KB)  PDF (2584 KB)  ( 128 )

658 Multi-disciplinary Citation Classification with Multiple Features Hot! Zheng Zhihan, Li Xinyu, Meng Fan, Bu Yi DOI: 10.3772/j.issn.1000-0135.2024.06.003 As the primary method to deeply understand citation behavior, citation classification research plays an important role in many scenarios, such as document management, retrieval, and utilization. This study uses machine learning methods to further explore citation classification by reviewing important citation behavior mechanisms and citation classification research. In this study, the fields of the original dataset can be supplemented and increased by matching the literature database and document analysis, and the features of four major categories that may be related to citation classification are extracted during the construction of the citation classification model. Thereafter, the feature selection is conducted using a simulated annealing algorithm. The results indicate that the established random forest model has the best performance on citation influence and citation function classification and outperforms the classification model combining the support vector machine with the SciBERT linear layer. The model established by the study improves the performance of automatic classification of multidisciplinary citations and the process of feature extraction and selection in research, as well as the exploration of the relationship between citation categories and some factors that have certain reference values for related research. 2024 Vol. 43 (6): 658-671 [Abstract] ( 101 ) HTML (154 KB)  PDF (1399 KB)  ( 102 )

672 Big Data Governance Mode Analysis Method for Urban Disaster Risk Response with Case Support Hot! Liu Zhaoge, Li Xiangyang, Qiao Limin, Wu Chong DOI: 10.3772/j.issn.1000-0135.2024.06.004 Big data applications encounter several problems, such as data imbalance, insufficient sharing, and a low level of precision and intelligence, in the field of urban disaster risk response (DRR). The construction and improvement of big data governance mode needs relevant knowledge support urgently; however, it faces the dilemma of lack of knowledge caused by scenario complexity and fragmented knowledge distribution. From the perspective of best practice theory, considering the knowledge learning of historical cases, this paper proposes a DRR big data governance mode analysis method based on case support. The core of the case-based method is to retrieve and transfer the governance modes that can be used for the target scenario through scenario similarity matching and build the available mode set. Targeting the retrieved available modes, the mode application effect data are combined to diagnose the mode application problems and select the governance modes. Subsequently, the integration of multi-case modes is completed from the perspective of management complexity and implementation cost to generate high-quality modes that can effectively solve the actual big data application problems. The rationality of the proposed analysis method is analyzed through a case study of urban community fire prevention in Puyang City, Henan Province. The use case results indicate that the proposed method can accurately transfer and apply historical experience, which is conducive to integrating fragmented knowledge, solving the problem of knowledge scarcity in complex scenarios, and constantly improving the value of big data application and governance mode in the DRR field. 2024 Vol. 43 (6): 672-684 [Abstract] ( 157 ) HTML (161 KB)  PDF (3167 KB)  ( 89 )

Intelligence Technology and Application

685 Identifying Core Users in Online Knowledge Community by Integrating Multiple User Attributes: Based on the Emotion-Weighted LeaderRank Algorithm Hot! Yang Ruixian, Yu Zhengjie, Zhong Qian, Liu Lili, Wei Huanan DOI: 10.3772/j.issn.1000-0135.2024.06.005 Based on the analysis of user attributes in network knowledge communities, a core user identification method that integrates multiple user attributes is proposed to improve the efficiency and effect of core user identification and provide theoretical and methodological reference for improving community operation and management levels. First, based on the user’s basic attribute data, the user’s activity and professionalism are quantified. Second, a hypernetwork model of the online knowledge community is constructed. An algorithm for the importance of user social relations based on the overlap of neighboring friends, a method for calculating cumulative interaction emotions in user interaction activities, and a ranking algorithm for user comprehensive emotional orientation are proposed. Finally, the entropy weight method is used to integrate the above indicators as the user’s core score, and core users are identified by sorting the scores. The results of empirical research indicate that, compared with the degree centrality ranking in the user social relationship network and the LeaderRank ranking in the user interaction relationship network, the method for identifying core users in the online knowledge community by integrating multiple attributes proposed in this study has better recognition effects. 2024 Vol. 43 (6): 685-696 [Abstract] ( 96 ) HTML (214 KB)  PDF (1547 KB)  ( 129 )

697 Potential High-Value Patent Identify Based on a Time-Series Graph Neural Network Hot! Zhou Xiao, Wang Bo, Hu Yulin, Wei Chuchu DOI: 10.3772/j.issn.1000-0135.2024.06.006 High-value patents are primary resources in constructing the current “dual circulation” development pattern at both domestic and international levels. They also play a pivotal role in positioning China at a strategic high ground in the new international economic order and comprehensively advancing technological self-reliance and self-strengthening. Precisely identifying potential high-value patents is a crucial step for nurturing their value and promoting technological transfer. Based on an in-depth analysis of the characteristics of patents that have won the China Patent Award, this study combines the use of Patent-BERT (bidirectional encoder representations from transformers for patent) and graph deep learning algorithms. By integrating patent evaluation indicators and textual features, we propose a potential high-value patent identification model based on graph convolutional networks (GCNs) and long short-term memory (LSTM) networks. The two main innovative aspects of this research are as follows: (1) Addressing the shortcomings of previous studies that only focused on “quantitative” features such as patent growth rate and collaboration potential and lacked deep semantic understanding of the text. We build a patent value representation model from both textual semantics and patent metrics dimensions. (2) Considering the temporal variability of patent value, we explore the evolutionary rules of patent value from a dynamic perspective, providing a new research approach for patent value mining and assessment. Finally, we compare the performance of various models, including node2vec, doc2vec, GCN, and multilayer perceptron (MLP). The results indicate that our model outperforms the control models across multiple indicators, thereby effectively validating the efficiency and robustness of our research approach. 2024 Vol. 43 (6): 697-711 [Abstract] ( 159 ) HTML (208 KB)  PDF (3952 KB)  ( 198 )

712 Identification of Problem and Method in Scientific Papers Based on Formulaic Expression Desensitization and Enhanced Boundary Recognition Hot! Zhang Yingyi, Zhang Chengzhi DOI: 10.3772/j.issn.1000-0135.2024.06.007 Problems and methods are crucial components of scientific papers and play a significant role in the organization, management, retrieval, and evaluation of scientific papers. To alleviate the formulaic expression dependency and word boundary recognition errors in problem and method recognition methods, we propose a model combined with formulaic expression desensitization and enhanced boundary recognition. Specifically, formulaic expression desensitization is achieved through data augmentation methods, whereas boundary enhancement utilizes pointer networks and sequence labeling models. With open access to scientific papers, researchers are utilizing full-text papers for entity recognition tasks. To demonstrate the importance of using full-text papers, this paper manually constructs an abstract and full-text annotated dataset in the field of natural language processing. Numerical and content-based metrics are designed to compare the problem, method, and their relationship extracted from two datasets. The results of ten-fold cross-validation experiments indicate that the proposed model outperforms baseline models such as SciBERT-BiLSTM-CRF significantly, with a macro-average F1 score improvement of 3.69 percentage points. When comparing entity recognition and relationship extraction results between abstracts and full texts, this paper shows that problem and method entities in abstracts have a broader semantic representation, whereas full texts contain more detailed entities and relationships that describe model design and training procedures. 2024 Vol. 43 (6): 712-732 [Abstract] ( 115 ) HTML (320 KB)  PDF (2519 KB)  ( 93 )

733 Evolutionary Game and Simulation Research on Competitive Intelligence Sharing Strategy among Small and Medium-Sized Enterprises Hot! Zhi Fengwen, Peng Zhaoqi, Zhao Mengfan, Zheng Yanning DOI: 10.3772/j.issn.1000-0135.2024.06.008 As the relationship among small and medium-sized enterprises (SMEs) shifts from competition to coopetition, the sharing of competitive intelligence among SMEs has become possible. This study aims to explore the evolutionary game process and influencing factors of competitive intelligence sharing strategies among SMEs. It is expected to provide a basis for the selection of SMEs’ sharing strategies and provide new ideas and references for subsequent theoretical research and practical exploration. Based on evolutionary game theory, this study deduces the evolutionary game process of competitive intelligence sharing between both sides, analyzes the gains and losses of benefits, the evolutionary stability strategy of the game system, and the degree of influence of different parameters, and verifies the results through numerical simulation. The results indicate that collaborative benefits and penalty costs have a significant positive impact on competitive intelligence sharing strategies among SMEs, while leakage risks and sharing costs have a significant negative impact on them. When all other parameters are determined, the quality of shared competitive intelligence and the absorption capabilities of competitive intelligence have a significant negative impact in a certain interval, and the smaller the gap among SMEs, the more likely they are to ultimately stabilize in the sharing strategy. Finally, targeted suggestions are proposed to promote competitive intelligence sharing among SMEs. 2024 Vol. 43 (6): 733-746 [Abstract] ( 128 ) HTML (256 KB)  PDF (2425 KB)  ( 76 )

Intelligence Reviews and Comments

747 Scientific Structure Review and Its Hierarchical System Construction Hot! Tian Qianfei, Chen Yunwei, Zhang Zhiqiang DOI: 10.3772/j.issn.1000-0135.2024.06.009 Scientific structure is the outline and framework of a system gradually accumulated by scientific knowledge after long-term development and change. Additionally, it is an internal relationship and logical mechanism that is not transferred by human will. By clarifying the concept and connotation of science and scientific structure, this paper comprehensively elucidates the multidisciplinary perspective research progress of “the research theoretical foundation of scientific structure laid by philosophy of science, deep excavation of the internal level of scientific structure by science of science, and research content of scientific economics to effectively expand scientific structure.” Based on the study of macro-, meso-, and micro-level systems of scientific structure and their interaction with society, a three-level system integration diagram of scientific structure with scientific knowledge as the core is constructed. In the 21st century, when global scientific and technological competition is unprecedentedly fierce, studying the scientific structure is conducive to revealing and grasping the development process and trend of science, as well as the global and national scientific development patterns, and plays a key role in assisting the formulation of appropriate science and technology policies and the establishment of reasonable scientific research institutions. 2024 Vol. 43 (6): 747-759 [Abstract] ( 126 ) HTML (142 KB)  PDF (2384 KB)  ( 103 )