Special Topics

1217 Big Data Support System for Science and Technology Innovation Policy Analysis Hot! He Defang, Zeng Jianxun, Chen Tao, Pan Yuntao, Yang Fangjuan DOI: 10.3772/j.issn.1000-0135.2025.10.001 Analysis of science and technology innovation policies covers the entire chain and processes of scientific and technological innovation activities. This analysis requires the utilization of big data resources, methodologies, technologies, platforms, and tools to support and empower the processes, models, and scenarios of science and technology innovation policy analysis. Building upon a clarification of the connotations of science and technology innovation policy analysis and its big data characteristics, this study expounds on the logical chain of big data to support this analysis from the perspectives of practical needs and new opportunities, the functional mechanism of big data in this context, and the framework for science and technology innovation policy analysis. Consequently, a big data support system tailored for the analysis of science and technology innovation policy is proposed, and the process of analysis based on big data is discussed on four levels: the big data resource system, refined data processing system, intelligent computing and analysis system, and evidence generation system. Finally, the study suggests continuous advancements in areas such as big data resource platforms, tool models, application environments, and theoretical methods to enhance policy analysis with support capabilities and resource infrastructure for big data evidence, methodologies, and simulation. 2025 Vol. 44 (10): 1217-1227 [Abstract] ( 60 ) HTML (89 KB)  PDF (1574 KB)  ( 44 )

Intelligence Theories and Methods

1228 Changes and Influence Mechanism of User Cognitive Level during Similar Public Emergency Situations Hot! An Lu, Wang Qinyan, Li Gang DOI: 10.3772/j.issn.1000-0135.2025.10.002 Studies on the changes in cognitive level of users in similar public emergency situations are required for emergency management departments to grasp the public's perception of the public events and its pattern as well as to design targeted emergency response measures. Based on the social cognition theory and distributed cognition theory, in this study, a method for measurement of user cognitive level is constructed, and changes in user cognitive level in similar public events are detected. The influencing factors of user cognitive level changes, as well as the importance, summary, and interaction of each factor, are analyzed. In the context of public events, users' cognitive levels vary largely. By participating in online public opinions of public events, users can improve their cognitive levels in subsequent similar public events. Relevant departments can promote the improvement of users' cognitive levels based on the revealed influence mechanism of changes in users’ cognitive levels. 2025 Vol. 44 (10): 1228-1241 [Abstract] ( 43 ) HTML (169 KB)  PDF (1744 KB)  ( 31 )

1242 Fine-Grained Sentiment Analysis of Online Government- Public Interaction Texts Using Large Language Models: Multistage Optimization Approach Hot! Teng Jie, He Huanglan, Hu Guangwei DOI: 10.3772/j.issn.1000-0135.2025.10.003 Online government-public interaction texts contain a rich emotional information that reflects public opinion, making the fine-grained sentiment analysis valuable for enhanced government governance capabilities. However, traditional methods struggle to accurately capture complex emotional features in these texts. In this study, we propose a multistage optimization framework for a fine-grained sentiment analysis of online government-public interaction texts based on large language models. First, we establish a sentiment classification system with eight emotional zones and 56 emotional labels based on the arousal-valence theory of emotion, and employ GPT-4 with BROKE framework-designed prompting strategies for an initial sentiment annotation, effectively addressing complex contextual understanding challenges. Second, we innovatively propose a “large model annotation-expert evaluation-judgment model training-multistage optimization” data quality control mechanism, using the Claude 3.5 Sonnet model for correction and judgment model for filtering, solving the “hallucination” problem of large models in domain applications. Finally, we achieve emotional category balancing through Claude 3.5 Sonnet’s data generation capabilities, reducing the imbalance Gini coefficient from 0.866 to 0.181, significantly enhancing the model generalization ability. Experimental results show that this framework achieves an accuracy of 89.70% in fine-grained emotional analysis tasks, with an F1 score improvement of 21.65% on average compared to existing methods, providing a new technical paradigm for an enhanced government-public interaction effectiveness. 2025 Vol. 44 (10): 1242-1258 [Abstract] ( 43 ) HTML (207 KB)  PDF (1766 KB)  ( 34 )

1259 Analysis of Textual Attributes of Scientific Tweets and Their Impact on the Cascading Evolution Trends of Scientific Papers Hot! Cao Renmeng, Xu Xiaoke, Wang Xianwen DOI: 10.3772/j.issn.1000-0135.2025.10.004 Scientific tweets are an important medium for the diffusion of scientific papers on social media. The understanding of the impact of textual attributes of scientific tweets on the diffusion effects of scientific papers helps science communicators in optimizing their strategies, broadening the reach of scientific information, and facilitating academic communication and public engagement. In this study, based on a dataset of over 50,000 papers and 400,000 scientific tweets, we explore the influence of various textual attributes on the cascade propagation dynamics of scientific tweets across three dimensions: tweet content, multimedia elements, and emojis. The results indicate that the use of the highlights of the paper as tweet content, along with incorporation of visual elements such as images, videos, and emojis into the tweets, can significantly enhance the diffusion scope of scientific papers. This effect not only is evident in the initial stages of propagation but also becomes more pronounced in subsequent stages, resulting in a “rich-get-richer” trend, i.e., Matthew effect in cascade diffusion. By introducing a computational communication perspective into altmetrics research, this study provides a deeper and more comprehensive understanding of the diffusion processes and patterns of scientific papers on social media, and thereby uncovers the mechanisms that drive the diffusion of scientific papers. 2025 Vol. 44 (10): 1259-1271 [Abstract] ( 33 ) HTML (121 KB)  PDF (2925 KB)  ( 50 )

Intelligence Technology and Application

1272 Local Citation Recommendation Based on Contextual Semantics and Global Information Hot! Zhang Xiaojuan, Ma Le DOI: 10.3772/j.issn.1000-0135.2025.10.005 Local citation recommendations can help researchers obtain relevant references efficiently and quickly. Existing local citation recommendation methods primarily focus on identifying candidate citations based on limited contextual semantic information, which generally provides limited improvements in the accuracy of the recommendation results. To improve the accuracy of local citation recommendations, this study further integrated global information (i.e., global semantics and global relations) based on contextual semantics. First, the pre-trained SciBERT (scientific bidirectional encoder representations from transformers) model was fine-tuned using custom tasks to extract the semantic embeddings of the citation context. To take advantage of the highly summarized global information provided by titles and abstracts of articles, we used Sentence-BERT and fine-tuned SciBERT models to extract embedded vectors for titles and abstracts of articles, respectively. Subsequently, a heterogeneous graph including three types of nodes—authors, papers, and venues (conference or journals)—was constructed, and a relational graph convolutional network (R-GCN) was used to aggregate three different types of relationships (citation, authorship, and publication) for generating embedded vectors for papers and authors. Finally, the issue of citation recommendation was formulated as a multi-classification task. The embedded representations of the citation context, title, abstract, target paper, and author of the target paper were concatenated to construct the input for the recommendation model. This model was then trained using a feedforward neural network (FFNN) and softmax to generate a list of candidate citations for a given context of a target paper. Two optimization strategies (distributed data parallelism on multiple devices and model compression on a single device) were leveraged to further improve the computational efficiency of the proposed model and reduce its operation time cost. Experimental results demonstrated that the proposed method effectively improves the accuracy of local citation recommendations. Based on contextual semantics, the embedded representations of the target paper contributed the most to the performance of the proposed model among all the vectors of global information (i.e., title semantic, abstract, author, and paper vectors). The contribution of global relations to the recommendation performance of the proposed model was greater than that of global semantics. Among all global relationships (citation, authorship, and publication), the historical citation relation contributed the most to improving the overall performance of local citation recommendations. Both optimization strategies improved the operational efficiency of the model. 2025 Vol. 44 (10): 1272-1286 [Abstract] ( 43 ) HTML (182 KB)  PDF (3575 KB)  ( 14 )

1287 Word Vector Network-based Analysis of Scientific Research Topic Evolution: Revealing the Semantic Drift Process Hot! Wang Hongyu, Shi Kaiwen, Wang Xiaoguang, Jin Zhuang, Zheng Yang, Huang Han DOI: 10.3772/j.issn.1000-0135.2025.10.006 As a classical knowledge network that characterizes the knowledge structure of discipline domains, the co-word network approach is affected by the sparse co-occurrence of feature keywords, different keyword synonyms, and insufficient corpus utilization. This makes it difficult to accurately depict the semantic correlations among keywords in the face of large-scale data in discipline domains. It is of practical significance to extend the co-word network from the theoretical and methodological levels to comprehensively reveal the semantic evolution process of discipline domain research topics at the macro and micro levels. This study considers the feature keywords of discipline domains as network vertices, obtains the vector representations of the feature keywords through the GloVe global word embedding model, and sets the normalized cosine similarity between the corresponding word vectors as concatenated edge weights to construct a fully-connected and undirected word vector network. Furthermore, this paper analyzes the roles and features of the discipline domain word vector networks, proposes a research topic semantic drift analysis framework based on word vector networks, and conducts a comparative analysis of the semantic association relations that it characterizes and co-occurrence relations in the co-word network. It is found that the proposed discipline domain word vector network, as a special class of knowledge network, is the mapping of the co-word network of featured keywords on the semantic hyperspace and has obvious value for the analysis of community structure and temporal evolution. Compared to the co-word network approach, the discipline domain word vector network is consistent in characterizing the key concepts of the discipline domain and is more stable and comprehensive in reflecting the knowledge structure of the discipline domain. It can reveal more detailed evolutionary processes, such as the semantic drift generated by scientific research topics at the micro level. 2025 Vol. 44 (10): 1287-1299 [Abstract] ( 36 ) HTML (181 KB)  PDF (1683 KB)  ( 51 )

1300 Sources of Basic Research for Disruptive Technologies: Individual Innovation Subject Perspective Hot! He Yubing, He Li, Xu Meijuan DOI: 10.3772/j.issn.1000-0135.2025.10.007 Disruptive technology, which is typically afforded by major breakthroughs in basic sciences, has been identified as a key factor that promotes scientific and technological revolution as well as industrial change. Thus, the basic research origins of disruptive technology must be examined to identify the core foundational disciplines and forecast future technological trajectories. In this context, individual-level innovation subjects is vital in bridging between basic research and disruptive technologies. Considering China’s graphene field as an example, this study investigates the knowledge linkage and group-knowledge heterogeneity between basic research and disruptive technologies from the perspective of individual innovation subjects. First, relevant innovation subjects are identified by screening disruptive technologies and relevant data from scientific papers are obtained. Second, topics are identified using the author-topic model, and the knowledge association between basic research and disruptive technologies is analyzed based on topic similarity. Finally, individual innovation subject groups are classified based on the number of disruptive technology inventions, and the research-interest diversity of low- and high-output innovation subject groups is analyzed using the topic-diversity index, while the author-topic association knowledge network is constructed to examine the group heterogeneity of basic research and disruptive technologies. The results indicate that scientific and technological topics in China’s graphene field have transitioned from a relatively superficial to a more profound development stage. A discernible knowledge link exists between basic research and disruptive technologies in individual innovation subjects, which has progressively strengthened over time. Individual innovation subjects who have invented a varying number of disruptive patents exhibit different topic diversities and extents of topic interconnectedness. Notably, highly productive inventors have gradually expanded their research depth while maintaining their research diversity. Thus, a distinctive research system characterized by a “small-world” effect within the group was developed. The findings of this study provide intelligence support for the identification, cultivation, and development of disruptive technologies as well as for the promotion of disruptive technology-oriented basic research. 2025 Vol. 44 (10): 1300-1314 [Abstract] ( 32 ) HTML (206 KB)  PDF (13098 KB)  ( 23 )

1315 Identification of Opportunities for Technological Fusion: Comprehensive Consideration of Relational and Structural Embedding Hot! Zhao Youlin, Gu Chenya, Wang Jiajie, Wang Yuemei, Shi Yanqing, Feng Li DOI: 10.3772/j.issn.1000-0135.2025.10.008 With the rapid development of technology and deepening of disciplinary integration, the technological innovations increasingly rely on the deep integration of interdisciplinary knowledge. In the existing research on identification of opportunities for technology integration, fields with high levels of technology integration have not been extensively investigated, and the adverse effects of weak relationships on knowledge fusion innovation in this field have not been considered. There is also a lack of consideration on the role of nodes in different network locations in technology fusion innovation. In this regard, this article introduces the theory of network embeddedness to construct a model that identifies opportunities for technology integration from fields with high levels of technology integration. The relationship embedding in the network embeddedness theory focuses on the strong and weak characteristics of relationships, while the structural embedding emphasizes the positional characteristics of nodes. First, the model introduced in this paper constructs a technology fusion path based on relationship embedding, dividing citation relationships into four quadrants with different strong and weak characteristics, combining them pairwise, and constructing knowledge networks separately. The knowledge network with a high technology fusion potential is used as an existing technology fusion path. Second, based on this technology fusion path and integration opportunity of structural embedding recognition technology, knowledge combinations that have not yet been connected and cross domain are identified through the medium of structural holes and central position nodes, and their fusion value is calculated to screen out technology fusion opportunities. An empirical evidence in the field of hyperspectral imaging with diverse fusion features shows that this model has a higher recognition accuracy and better depth and breadth of recognition content, which helps grasp the direction of future technological fusion. 2025 Vol. 44 (10): 1315-1328 [Abstract] ( 25 ) HTML (167 KB)  PDF (1503 KB)  ( 30 )

Intelligence Discipline Development and Construction

1329 Challenges and Feasible Measures for Long-term Development of Information-Science Faculty in New Era Hot! Wang Wei, Yang Jianlin, Liang Jiwen DOI: 10.3772/j.issn.1000-0135.2025.10.009 As a vital component of information-resource management, the development of information-science faculty, which is under the context of national strategic transformation and the innovation of digital and intelligent technologies, is confronted with new challenges and opportunities in the new era. By reviewing the historical evolution of faculty development, some issues are identified in faculty construction that pertain to talent evaluation, the positioning of the discipline and faculty roles, comprehensive capacity, intelligence research and education, and practical capabilities. This study proposes adjustment to the role positioning of faculty while considering major national strategies, followed by clarifying the future vision of faculty in terms of comprehensive literacy, discipline construction, academic research, practical applications, and professional development. Focusing on the synergistic fusion among talent, industrial, and innovation chains, feasible measures are identified for faculty construction, including enhancing intelligence literacy, applying intelligence theories and methods to industrial security and development, and establishing a digital support system to facilitate faculty growth. These strategies are anticipated to provide long-term support for the sustainable development of the information-science faculty. 2025 Vol. 44 (10): 1329-1341 [Abstract] ( 33 ) HTML (126 KB)  PDF (2986 KB)  ( 25 )

Intelligence Reviews and Comments

1342 Review of Weak Signal Identification Based on Technology Foresight Perspective Hot! Cao Haiyan, Wang Nuanchen, Mu Ge, Li Wanhong DOI: 10.3772/j.issn.1000-0135.2025.10.010 Weak signals serve as early indicators of future technological opportunities and are crucial for innovation risk mitigation. Thus, they hold significant value in seizing strategic initiatives. However, current academic research results on how to identify weak signals are somewhat scattered. Especially, weak signal identification based on the perspective of technology foresight faces problems such as overlapping research fields and fragmented knowledge, thereby lacking effective support for predicting future technological development trends. In this paper, through a systematic literature review, combined with the results of keyword clustering and weak signal identification processes, a three-layer research framework was constructed, consisting of trend monitoring, signal cognition, and value construction layers. First, the themes, fields, and data of weak signal identification in the trend-monitoring layer were analyzed. Second, the characteristics and methods of weak-signal identification in the signal cognition layer were deconstructed. Third, the representation, verification, and interpretation of weak-signal identification results in the value-construction layer were sorted. Finally, by integrating the research results of the three layers, the overall framework of weak signal identification was constructed based on the perspective of technology foresight. The outlook for future research is proposed accordingly, with the aim of providing theoretical and practical references for exploring future technological opportunities and promoting the development of emerging industries by clarifying the research work on weak signal identification from the perspective of technology foresight. 2025 Vol. 44 (10): 1342-1358 [Abstract] ( 47 ) HTML (209 KB)  PDF (2353 KB)  ( 39 )