Intelligence Theories and Methods

889 Research on Motives and Approaches for the Expansion of Intellectual Property Intelligence Theories Hot! Li Jianfei, Shen Shujing, Yang Jianlin DOI: 10.3772/j.issn.1000-0135.2024.08.001 Intellectual property intelligence is an important component of scientific and technological intelligence, providing theoretical support for intellectual property intelligence work. With the increasingly complex international competitive environment, intellectual property intelligence has become increasingly prominent in supporting national security and the development of the technology industry. Its goal positioning and research innovation in the development of China’s intelligence studies require specialized theoretical support. This article is based on a review of the research on intellectual property intelligence theories, combined with the requirements of intelligence work practice and disciplinary development, to explore the motives and approaches for the expansion of intellectual property intelligence theories. According to the research, national strategy guidance, practical demand drivers, academic proposition condensation, and discipline paradigm transformation are the motives for expanding on the intellectual property intelligence theories; furthermore, clarifying the research questions, deepening the basic theories, building the theoretical systems, and improving the academic discourse are the most suitable approaches to expanding on intellectual property intelligence theories. 2024 Vol. 43 (8): 889-898 [Abstract] ( 60 ) HTML (102 KB)  PDF (761 KB)  ( 104 )

899 Identification of Disruptive Technologies and Prediction of Diffusion Trends: Conceptual Model and Empirical Analysis Hot! Wang Kang, Chen Yue, Wang Yuqi, Han Meng DOI: 10.3772/j.issn.1000-0135.2024.08.002 Identifying and assessing the disruptive potential and diffusion trends of a technology can provide a basis for precise decision-making regarding the allocation of scientific and technological resources and advanced layout of future industries for the country and government. This study first constructs a conceptual model for identifying disruptive technologies and predicting their diffusion trends. Subsequently, based on this model, using the 3D printing field as an example, disruptive patents are identified from among outliers and influence dimensions, and disruptive technologies are extracted. Finally, based on the identification of disruptive patents, automatic labeling and strategic coordinates are applied to technology topic diffusion paths. A new multistate automatic labeling technology topic diffusion trend prediction method is proposed to reveal the dynamic diffusion relationship between core, edge, mature, emerging, and other positional themes. An inherent logical relationship among symbiosis, matching, and the association between outlier patents and disruptive technologies is established through further research. Thus, identifying disruptive technologies from the perspective of outlier patents is feasible. The disruptive technologies in the field of 3D printing from 1955 to 2017 were mainly distributed in three directions: high-end equipment manufacturing, biopharmaceuticals, and materials. The prominent technological fields were transportation, engines/pumps/turbines, biomaterial analysis, semiconductors, and environmental technology. The predicted trend of topic diffusion in multi-state automatic labeling technology indicates that the future development potential of biomedical 3D printing technology is enormous. 2024 Vol. 43 (8): 899-913 [Abstract] ( 59 ) HTML (120 KB)  PDF (4776 KB)  ( 114 )

914 Research on Infodemic Management Capacity Hot! Pei Junliang, Zhou Xiaoying DOI: 10.3772/j.issn.1000-0135.2024.08.003 Infodemic management requires adequate capacity, but the current research on infodemic topics rarely involves capacity issues. Based on an analysis of the infodemic management capacity-building measures carried out by the World Health Organization, this paper studies the content on infodemic management capacity and establishes a model for ensuring adequate infodemic management capacity to fully reflect the basic capabilities that infodemic managers must master in the future and provide a reference and standards for the education and training of infodemic managers in China. In this paper, the network survey method is used to collect the relevant data on the infodemic management capacity-building measures carried out by the World Health Organization, including capacity training and capacity framework. The qualitative research method of semi-naturalism is adopted, and the steps of “discretization processing of materials - clustering processing of items - relationship analysis of clusters” are followed to conduct a systematic analysis. Based on a systematic analysis of the materials, a model of infodemic management capacity is constructed, which consists of six components: infodemic cognition ability, infodemic monitoring ability, infodemic analysis ability, infodemic intervention ability, infodemic evaluation ability, and supportive ability, and the specific content of the six abilities is studied. To ensure adequate infodemic management capacity, it is necessary to adhere to the strategies, which include a combination of inter-disciplinary and inter-system measures, theory and practice, stability and iteration, emergency management and sustainability, and full-cycle and all-round measures. 2024 Vol. 43 (8): 914-926 [Abstract] ( 46 ) HTML (148 KB)  PDF (1687 KB)  ( 48 )

927 Co-occurrence Mapping between Chinese Library Classification and International Patent Classification Categories Based on Author's Research Similarity Hot! Song Yanhui, Chen Xinqi DOI: 10.3772/j.issn.1000-0135.2024.08.004 Knowledge flow between papers and patents reflects the evolutionary route of scientific research and technological innovation. Category mapping analysis of Chinese Library Classification (CLC) and International Patent Classification (IPC) is helpful in breaking through the barriers between paper and patent resources, by identifying the characteristics of scientific and technological development between papers and patents of different disciplines. This study proposes a mapping method that integrates the idea of social network analysis with co-occurrence mapping. Taking the papers and patents of the same author and highly related research topics as unit data, CLC and IPC are used to classify and label each unit of data simultaneously, thereby combining category similarity calculations and analyzing the labeling results of the dataset. Finally, universal one-to-one, one-to-many, and two-way mappings between the CLC and IPC categories are obtained. 2024 Vol. 43 (8): 927-935 [Abstract] ( 27 ) HTML (116 KB)  PDF (1925 KB)  ( 75 )

936 Research on Privacy Data Identification and Measurement Based on Medical Information Text Hot! Zhang Kailiang, Zang Guoquan, Xiao Yang DOI: 10.3772/j.issn.1000-0135.2024.08.005 The results of data classification in medical industry standards are fuzzy, with few accompanying measurement results. Considering existing problems, this study adopted medical information text mining to objectively measure medical data privacy. Measurement results can provide a reference for verifying and improving current medical data classification results. In this study, the sources of medically sensitive data included industry standards, legal regulations, academic papers, and breach cases. The medically sensitive data unit is composed of sensitive nouns (also known as sensitive data items), sensitive verbs, and sensitive degree words, which are used in the privacy recognition model. The privacy measurement model considers the sensitivity, semantic strength, and text strength of sensitive data. In ranking the results of privacy values, medical application data ranked the highest, followed by health status, medical payment, and personal attribute data. 2024 Vol. 43 (8): 936-945 [Abstract] ( 46 ) HTML (123 KB)  PDF (1134 KB)  ( 91 )

Intelligence Technology and Application

946 Large Language Model-Driven Academic Text Mining: Construction and Evaluation of Inference-End Prompting Strategy Hot! Lu Wei, Liu Yinpeng, Shi Xiang, Liu Jiawei, Cheng Qikai, Huang Yong, Wang Lei DOI: 10.3772/j.issn.1000-0135.2024.08.006 Task comprehension and instruction-following abilities of large language models enable users to complete complex information-processing tasks through simple interactive instructions. Scientific literature analysts are actively exploring the application of large language models; however, a systematic study of the capability boundaries of large models has not yet been conducted. Focusing on academic text mining, this study designs inference-end prompting strategies and establishes a comprehensive evaluation framework for large language model-driven academic text mining, encompassing text classification, information extraction, text reasoning, and text generation, covering six tasks in total. Mainstream instruction-tuned models were selected for the experiments, to compare the different prompting strategies and professional capabilities of the models. The experiments indicate that complex instruction strategies, such as few-shot and chain-of-thought, are not effective in classification tasks, but perform well in more challenging tasks, such as extraction and generation, whereby trillion-parameter scale models achieve results comparable to those of fully trained deep-learning models. However, for models with billions or tens of billions of parameter scales, there is a clear upper limit to inference-end instruction strategies. Achieving deep integration of large language models into the field of scientific intelligence requires adaption of the model to the domain at the tuning end. 2024 Vol. 43 (8): 946-959 [Abstract] ( 66 ) HTML (159 KB)  PDF (3212 KB)  ( 185 )

960 Research on Constructing an Academic Knowledge Graph of Multi-dimensional Knowledge Elements in Academic Full Texts Hot! Shen Si, Zhu Yufei DOI: 10.3772/j.issn.1000-0135.2024.08.007 Academic texts contain a large amount of knowledge element information. Mining and organizing these knowledge elements can effectively improve the utilization efficiency of academic resources. Through the construction of an academic knowledge graph, connecting all kinds of tacit “knowledge elements” in an article can not only save time for scholars seeking to obtain knowledge points but also help them expand knowledge points through the network community in a knowledge graph. Through literature research and other methods, beginning with three dimensions, this paper determines the key knowledge elements in 18 academic papers, takes the text description information of knowledge elements as the entity object, and outlines the conceptual framework of an academic knowledge graph. Then, 515 pieces of literature in the JASIST are selected to study the manual annotation and entity extraction of the key knowledge elements in each paper based on deep learning. The research content includes whether such knowledge elements will create problems in the process of manual annotation and whether they will reach the expected value in automatic extraction when attempting to screen the knowledge elements involved in the construction of a knowledge graph. Finally, nine kinds of knowledge elements are selected, including mathematical formulas, software tools, data sources, specific models, tables, graphs, research prospects, research problems, and research results. Together with the titular data, triads composed of head entities, relations, and tail entities are generated and stored in the graph database for visual evaluation. Finally, the visualization and knowledge element retrieval of the graph are studied to prove its feasibility and scalability. The research shows that some knowledge elements in the text are suitable for large-scale automatic annotation, and all kinds of knowledge elements can form a dense knowledge community through mutual links. 2024 Vol. 43 (8): 960-975 [Abstract] ( 62 ) HTML (150 KB)  PDF (3067 KB)  ( 65 )

976 Interdisciplinary Measurement Research Based on Reference Literature and Text Content Subject Classification Hot! Lyu Qi, Shangguan Yanhong, Li Rui DOI: 10.3772/j.issn.1000-0135.2024.08.008 Currently, the world is facing several scientific challenges that require collaborative efforts from multiple fields and disciplines to solve them in various ways. Therefore, grasping the development trend of interdisciplinary integration and exploring the degree of interdisciplinary research in different disciplines is an important research direction. This paper combines reference and text content information to construct a citation embedding SCIBERT attention model, which classifies journal literature by discipline and conducts interdisciplinary measurement research based on the results of discipline classification. Simultaneously, the feasibility and effectiveness of the method proposed in this study are verified by comparing it with interdisciplinary measures of single input references and text content. Research has found that integrating text content and reference information can effectively achieve disciplinary classification of journal literature, which is better than disciplinary classification with a single input of information. A disciplinary classification based on references and text content can effectively conduct interdisciplinary measurement. The granularity of the disciplinary classification system will affect the disciplinary classification effectiveness of journal literature. 2024 Vol. 43 (8): 976-991 [Abstract] ( 39 ) HTML (181 KB)  PDF (2224 KB)  ( 101 )

992 Patent Infringement Risk Early Warning: Using Knowledge Graphs Hot! Ding Shengchun, Qin Tianyun, Wang Yilin DOI: 10.3772/j.issn.1000-0135.2024.08.009 Patented technologies for products in the same field have characteristics such as high technical relevance and enterprises face the potential danger of patent infringement in their operations and production activities. Based on the actual needs of patent infringement early warning of enterprises, it is of great significance to efficiently and accurately detect the risk of patent infringement of products; thus, this study proposes a Patent Infringement Risk Early Warning Model. In this model, the schema layer of the domain patent knowledge graph and product knowledge graph are redefined, covering three types of entities: component entities, structural entities, and efficacy entities, as well as four types of entity relationships: composition relationship, relative position relationship, connection relationship, and efficacy achievement relationship. Patent and product technical scheme knowledge graphs are constructed based on the BERT and BiLSTM models. Based on the ComplEx model, the knowledge graph was embedded to achieve a quantitative calculation of the similarity between products and patented technologies, and an infringement warning was issued according to the patent infringement risk index. Two types of products, air humidifiers and earphones, were used for the empirical research, and the accuracy rate of patent infringement warnings was 86.67%, which has a certain application value. 2024 Vol. 43 (8): 992-1002 [Abstract] ( 53 ) HTML (106 KB)  PDF (2211 KB)  ( 86 )