Special Topics

1 Research on the Information Analysis Pattern Based on Large Models Hot! Lu Xiaobin, Zhang Zhouwentao, Huo Chaoguang DOI: 10.3772/j.issn.1000-0135.2026.01.001 Amid the new wave of scientific and technological revolution and industrial transformation, large models—which serve as a new foundational infrastructure in the digital intelligence era—are significantly reshaping the logic of information, knowledge production, and value creation. As a secondary discipline in information resource management that relies heavily on techniques and methods, information analysis urgently needs to transform working patterns based on this emerging technological foundation. Accordingly, based on a review of the mechanisms and developmental landscapes of large models, this study systematically examines their major impacts on existing information analysis workflows. It emphasizes the transformation of intelligent information analysis processes from the perspective of large models. Four intelligent information analysis working patterns are identified: prompt engineering, fine-tuning, continued pretraining, and domain-specific autonomous pretraining enhancement. For each pattern, this study elaborates on its operating principles, identifies applicable information analysis task scenarios, and presents the corresponding technical routes for intelligent information analysis. As the first theoretical investigation on large-model-driven intelligent information analysis, this study remodels the conventional methodological framework, establishes four new large-model-based working patterns, and provides new methodological support for information and intelligence analysis, management decision-making, and related applications in China. 2026 Vol. 45 (1): 1-18 [Abstract] ( 0 ) HTML (166 KB)  PDF (7052 KB)  ( 18 )

Intelligence Theories and Methods

19 Prediction and Selection of Potential Transformation Paths for Scientific and Technological Achievements Based on Knowledge Flow Multilayer Network Hot! Dong Kun, Chen Kexin, Zhang Xiaohui, Mu Qingxiu, Cui Zihang, Cui Bin DOI: 10.3772/j.issn.1000-0135.2026.01.002 The scientific and technological achievements in China are transformed inefficiently. Therefore, this study constructed a methodological framework based on a knowledge flow multilayer network to predict and select potential pathways of achievement transformation, thereby strategically promoting the flow of scientific knowledge along the entire chain of “scientific research—technological development—industrial application.” First, a knowledge flow multilayer network integrating multivariate relationships—including thematic citation, thematic application, and collaborative innovation and transfer among entities—was established using data from academic papers, patents, and product information. Subsequently, link prediction methods were employed to analyze potential transformation pathways, and an evaluation system was developed to prioritize these pathways based on three dimensions: path maturity, conversion efficiency, and implementation difficulty. Finally, an empirical study was conducted in the vaccine field. Through comparative analyses and case retrospective validation, the effectiveness and feasibility of the method were confirmed. The results demonstrate that this multilayer network can comprehensively characterize the complex knowledge flow in the transformation of scientific and technological achievements. Combined with the analytical framework integrating link prediction and multi-dimensional metrics, it enables accurate identification and optimization of potential pathways, with the automated processes significantly enhancing analytical efficiency. However, in practical applications, challenges such as coarse granularity of topic association, high difficulty in integrating multi-source data, and insufficient adaptability to industrial scenarios still need to be addressed. 2026 Vol. 45 (1): 19-35 [Abstract] ( 0 ) HTML (180 KB)  PDF (7152 KB)  ( 14 )

36 Recommendation Methodology for Enterprise Cooperation Partners Based on Cross-Attention Multi-head Contrastive Network Hot! Duan Yongkang, Zhao Guangyu, Geng Qian, Jin Jian DOI: 10.3772/j.issn.1000-0135.2026.01.003 The traditional enterprise partner recommendation method overly relies on technical characteristics and ignores the influence of multidimensional factors. Therefore, this study aimed to explore the multidimensional influence factors and prediction mechanism of enterprise partnership, to provide technical support for enterprises in finding suitable partners and formulating effective innovation strategies. The proposed cross-attention multi-head contrastive network (CAMC-Net)-based enterprise partner recommendation method that fuses enterprise, patent and policy data, models the bidirectional complementary characteristics of enterprise relationships through a cross-multiple attention mechanism, and introduces a comparative learning strategy to optimize the spatial distribution of enterprise representations. Taking the new energy industry as an example, during validation on a corporate collaboration dataset with International Patent Classification (IPC) H02P and H10, the CAMC-Net model achieved area under the receiver operating characteristic curve (AUC) scores of 0.9425 and 0.9251, accuracy rates of 0.8644 and 0.8387, and F1 scores of 0.8707 and 0.8471, respectively, in corporate relationship identification tasks, outperforming baseline models. Ablation experiments validated the effectiveness of policy data and model components. However, current research data primarily rely on a single domain, requiring future exploration of cross-domain enterprise partner recommendation methods. Additionally, the model does not consider multimodal data, necessitating further investigation into more efficient multimodal feature fusion strategies. 2026 Vol. 45 (1): 36-50 [Abstract] ( 0 ) HTML (205 KB)  PDF (3364 KB)  ( 8 )

51 Researcher Mobility Network and Academic Performance of Universities: A Study Based on Global Resume Data from ORCID Hot! Hu Xinyue, Lin Ziyang, Tang Yongli DOI: 10.3772/j.issn.1000-0135.2026.01.004 To accelerate the construction of the world's important talent center and innovation highland, it is particularly important for China to occupy a favorable position in the global talent mobility network. As a major concentrated hub for researchers and a key innovation entity, universities play a crucial role in this process. Existing literature on inter-organizational mobility networks primarily focuses on firms as the research unit and fails to examine the impact of the social networks formed by the inter-university mobility of researchers on organizational performance. This study leverages large-scale resume data from approximately 2 million researchers in ORCID to construct a global researcher mobility network encompassing over 15,000 universities worldwide. An empirical analysis is conducted to examine the relationship between the position of universities within the researcher mobility network and their academic performance. The results show that the in-degree centrality of the researcher mobility network significantly enhances both total and high-quality research output, whereas outdegree centrality exhibits an inverted U-shaped relationship with both total and high-quality research output. This study provides significant implications for talent management mechanisms of universities and talent policy of governments. 2026 Vol. 45 (1): 51-68 [Abstract] ( 0 ) HTML (287 KB)  PDF (3523 KB)  ( 6 )

69 Differences Between Artificial Intelligence and User-Generated Contents in Public Emergency Public Opinion Events— Based on Thematic and Emotional Perspectives Hot! Zhang Liu, Bi Hefei, Huang Bo, Xiang Mengmeng DOI: 10.3772/j.issn.1000-0135.2026.01.005 The development of generative artificial intelligence is profoundly changing the pattern of information dissemination in public opinion events. In the era of digital intelligence, the coexistence and integration of artificial intelligence of generative content (AIGC) and user-generated content (UGC) bring new opportunities and challenges to information governance in social media. The purpose of this study was to explore the differences between AIGC and UGC in theme characteristics and emotional expression in public opinion emergencies, to tap their complementary advantages and help in the development of a positive public opinion environment. This study extracted topic features using the bidirectional encoder representations from transformers (BERTopic) model, and fused self-attention mechanism with the bidirectional long short-term memory (Bi-LSTM) to extract emotion features from UGC comment data on the topic of “Capricorn Typhoon” on Weibo and AIGC data generated by ChatGPT and DeepSeek. The analysis was conducted from the perspectives of topic and theme-emotion fusion. The results show that AIGC has advantages in topic distribution novelty and coherence of document hierarchical structure, excelling in the mining and systematic expression of professional domain knowledge. UGC is cohesive and resonates more with social issues, with subject words in social context having unique meanings. AIGC mainly focuses on neutral emotions with a balanced distribution, information provision, and rational analysis, whereas UGC expresses rich and strong emotions and is more likely to trigger public resonance and social participation. In the future, the complementary advantages of AIGC and UGC can be employed in the management of public opinion emergencies. With AIGC improving the ability of semantic analysis and emotional expression and UGC capturing the true emotions of the public and promoting diversified dissemination of information, they jointly build a harmonious public opinion ecology. 2026 Vol. 45 (1): 69-87 [Abstract] ( 0 ) HTML (153 KB)  PDF (11173 KB)  ( 7 )

Intelligence Technology and Application

88 Knowledge Computing Framework and System Implementation for Holistic Policy Text Mining Hot! Ba Zhichao, Zhang Yujie, Liu Leilei, Meng Kai DOI: 10.3772/j.issn.1000-0135.2026.01.006 Data-driven paradigms in large-scale policy text mining are pivotal for enhancing governance effectiveness and optimizing decision-making processes. However, the intelligentization of policy analysis still faces numerous unresolved challenges when confronted with complex, dynamic decision contexts and practical demands. To address the shortcomings in existing policy research, particularly the lack of holistic consideration and insufficient cross-domain integration capabilities, this study explores a knowledge computing framework, its key enabling technologies, and system implementation, for holistic policy text mining. First, to clarify the concept of holistic policy, a framework is presented for policy knowledge computing, establishing a comprehensive lifecycle that integrates policy data acquisition and parsing, data fusion and association, and knowledge computation and reasoning. Second, the study dissects the critical technological pathways within the knowledge computing framework, to propose implementation strategies for fine-grained content parsing and linking, multi-dimensional policy element identification and automated comparison, policy citation detection and diffusion trajectory measurement, along with traceability analysis through multi-agent and policy network collaboration. Finally, leveraging a national corpus of over 1.58 million full-text policy documents, a policy knowledge computing system was constructed as a practical exemplar. This system realizes interconnected pathways across content, knowledge, and application layers, offering empirical insights and providing practical references for intelligent policy applications. 2026 Vol. 45 (1): 88-102 [Abstract] ( 0 ) HTML (137 KB)  PDF (6492 KB)  ( 18 )

103 Predicting Unlocking Potential of Technology Trajectories Based on Innovations of Principle and Application Hot! Chen Wen, Ma Yaxue, Ba Zhichao, Li Gang DOI: 10.3772/j.issn.1000-0135.2026.01.007 In the context of breakthrough innovations or paradigm shifts, technologies can unlock their original trajectories and transition to higher ones. This study focuses on the phenomenon of technology trajectory unlocking and explores the impact of technological principles and application innovations in this process. First, technology trajectories were constructed based on the cumulative distribution of patents, with the transition point to a new trajectory defined as the unlocking point. Indicators for technological principles and application innovation characteristics were then developed across four dimensions: innovation impact, frontier, novelty, and complexity. This study analyzed the differences in these indicators for technologies that experienced unlocking and those that did not. Finally, using these characteristics as input variables, machine learning algorithms were employed to predict the unlocking potential. Empirical research in the field of photolithography has revealed significant differences in technological principles and application innovations between unlocking and non-unlocking technologies, indicating that these characteristics can effectively predict the unlocking potential. The study enhances the understanding of technological evolution patterns and assists enterprises and researchers in identifying development potentials and capturing technological opportunities. 2026 Vol. 45 (1): 103-115 [Abstract] ( 0 ) HTML (195 KB)  PDF (2687 KB)  ( 9 )

Intelligence Users and Behavior

116 The Empowerment Effects of Government Information Design Based on Public Perception of Crisis Situations Hot! Fan Sinuo, Song Jinhui, Li Yuelin DOI: 10.3772/j.issn.1000-0135.2026.01.008 In the context of public crisis, improving the mechanism of government information disclosure is a key pillar of crisis management. Governments should empower the public in different crisis situations through appropriate information design. This study developed a theoretical model based on the Construal Level Theory and Empowerment Theory. A questionnaire survey was conducted to collect data from residents who experienced a flood in 2023. A structural equation model and fsQCA were used to analyze the data. The findings reveal that the psychological distance positively affects cognitive and emotional empowerment. Although information design with emotional appeal significantly enhances perceived empowerment, its mediating effect is not supported. Perceived severity and crisis emotions, such as sadness, fear, hate, and anxiety, negatively moderate the relationship between psychological distance and perceived empowerment. Through a fuzzy-set qualitative comparative analysis, this study further identified that information design can enhance the empowering effect of government information disclosure through emotion-centered, rational-centered, and emotion-rationality-integrated methods. Each is suited to different types of perceived crisis situations and public psychological states. This study discusses the empowerment mechanism of government information disclosure from three perspectives: informal information sources, objective crisis situations, and a combination of perceived crisis situations and public psychological states. This reveals the complexity of information disclosure during crises and provides practical guidance for government information disclosure and communication strategies. 2026 Vol. 45 (1): 116-130 [Abstract] ( 0 ) HTML (192 KB)  PDF (1447 KB)  ( 12 )

Intelligence Reviews and Comments

131 Comprehensive Review on Content-based Measurement of Academic Innovation Hot! Chen Danlei, Hua Bolin DOI: 10.3772/j.issn.1000-0135.2026.01.009 Academic innovation is a multi-dimensional and multi-faceted concept. Adopting quantitative methods to evaluate the innovative value of academic texts comprehensively, objectively, and rationally is vital in sustaining the dynamism of the academic ecosystem and promoting the deep implementation of the innovation-driven development strategy. This study maps the theoretical landscape of academic innovation and develops a fine-grained taxonomy of methods that measure academic innovation based on textual content. Within the proposed framework, we systematically review existing efforts and identify 10 technical approaches that either rely solely on textual content or integrate textual and external features to quantify academic innovation. Specifically, the text-only approaches include (1) analyzing the focal paper by mining self-assessment and peer-evaluation statements or calculating the distance between internal knowledge elements; and (2) comparing the focal paper with relevant studies by measuring its divergence from existing knowledge, deviation from mainstream research, or coverage of emerging topics. The hybrid approaches involve two strategies: (1) fusing multiple indicators through weighted rules, coordinate-based ranking, or unit conversion; and (2) reformulating innovation measurements as proxy tasks to enable in-model feature fusion via concatenated vectors or heterogeneous graphs. Subsequently, we utilize content analysis to synthesize prior researches in terms of feature types, textual granularity, empirical samples and their disciplinary scopes, as well as validation methodologies. Furthermore, we analyze the correspondence between the conceptualization and operationalization of academic innovation. Despite recent advancements, several critical challenges remain unresolved, such as the limited semantic roles of knowledge entities, insufficient sentence-level quantitative indicators, inadequate high-quality benchmark datasets, and difficulties in horizontal comparisons among state-of-the-art metrics. Building on these findings, this study outlines promising directions for further exploration. 2026 Vol. 45 (1): 131-149 [Abstract] ( 0 ) HTML (255 KB)  PDF (2418 KB)  ( 5 )

150 Human-Technology Coordination: Nature, Theoretical Framework, and Prospects of Digital Elderly-Orientation Hot! Wu Jiang, Chen Nan, Yuan Yiming, Zhao Yuxiang DOI: 10.3772/j.issn.1000-0135.2026.01.010 Against the dual backdrop of digitalization and population aging, the digital elderly-orientation has emerged as a critical component of people’s aspirations for a better quality of life and an increasingly important requirement for the development of digital technologies. As emerging technologies continue to evolve rapidly, individuals’ capacity to adapt to digital technologies plays a growing role in shaping both their self-perception of aging and the broader social evaluation of old age. These dynamics underscore the need to further clarify and critically examine the conceptual core and underlying logic of digital elderly-orientation to support the high-quality development of aging societies. This study first delineates the conceptual boundaries of “aging” within the context of digital elderly-orientation, arguing that its target population extends beyond older adults to include broader aging-oriented groups. It then examines the fundamental role of human-technology coordination as the core mechanism underlying digital elderly-orientation, thereby establishing its theoretical lineage and academic positioning. Building on an analysis of the human-technology origins and normative rationale, the study systematically reviews the major research themes in this field and develops an integrated theoretical framework grounded in sociotechnical systems theory. From the disciplinary perspective of information resources management, the discussion focuses on refined segmentation of aging-oriented populations and their information needs, the development of intelligent elderly-orientation products and information service systems, and the optimization of system-level policy intelligence and performance evaluation. Collectively, this study provides theoretical clarification and practical guidance for the continued advancement of digital elderly-orientation. 2026 Vol. 45 (1): 150-164 [Abstract] ( 0 ) HTML (176 KB)  PDF (2882 KB)  ( 5 )