Research on Emerging Technology Trend Prediction and Innovation Insight in the Rare Earth Industry Based on Patent Data Management and Knowledge Intelligence = 특허 데이터 관리 및 지식 인공지능 발견을 기반으로 한 희토류 산업의 신흥 기술 동향 예측 및 혁신 인사이트 연구

국문 초록 (Abstract)

세계 기술 혁신의 배경 속에서, 특허 데이터를 기반으로 한 기술 관리 및 지식 발견은 기술 발전과 경쟁력 향상을 위한 핵심 전략적 요구 사항이 되었다. 최근의 국제 연구 동향 및 정책 보고서는 미래 발전에 미칠 신흥 기술의 영향을 강조하고 있다. 예를 들어, 2024 년 세계 경제 포럼이 발표한 ‘10 대 신흥 기술 보고서’ 에서는 합성 생물학, AI 기반 화학, 양자 컴퓨팅 등이 향후 5 년 내에 중요한 경제적 및 사회적 영향을 미칠 것이라고 지적하였다. 마찬가지로, 맥킨지의 ‘2024 기술 트렌드 전망 ’에서는 글로벌 탈탄소화와 에너지 안보에서 생성형 AI 와 적용 AI 의 중요성을 강조하고 있다. 이러한 국제적 관점은 신흥 기술이 세계 정책과 산업 계획에서 차지하는 중요한 위치를 보여주며, 기술 혁신 연구에 대한 강력한 참고 자료를 제공하였다. 신흥 기술의 식별, 추적, 예측 및 관리 문제는 기술 혁신 시스템의 중요한 부분으로 오랫동안 주목받아 왔다. 신흥 기술을 정확하게 추적하고 예측하는 것은 기술 혁신과 자립, 국가 기술 독립을 추진하는 핵심 요소가 되었다. 특허 데이터는 기술 개발 및 혁신의 지적 정수로서, 기술 혁신, 전략적 기획 및 신흥 기술 식별에 중요한 역할을 담당하고 있다. 따라서, 특허 데이터를 기반으로 한 기술 관리와 지식 발견은 기술 혁신에서 실용적이고 과학적인 필수 요소로 자리잡았다. 본 논문은 특허 데이터를 심층적으로 탐구하고 신흥 기술의 상호 관계를 분석하였으며, 특허 요소와 신흥 기술의 특성에 중점을 둔 특허 데이터를 기반으로 한 기술 관리 및 지식 발견 모델을 구축하였다. 지식 그래프를 도입하여 대규모 데이터, 데이터 관계 및 지식 조직을 이해하는 데 있어 그 이점을 활용하여, 특허 데이터를 체계적으로 분석하고 조직화하며, 특허 데이터를 기술 지식으로 전환하는 데 기여하였다. 본 논문의 연구는 문헌 분석, 표현 학습, 기계 학습 및 실증 연구를 활용하여 특허 지식 그래프를 기반으로 한 기술 식별, 진화 및 예측 모델을 구축하고, 신흥 기술 추적 및 예측을 위한 기술 관리 및 지식 발견의 메커니즘과 전략을 제안하여 국가 기술 혁신 전략을 지원하였다. 논문에서 제 3 장은 특허 데이터의 구성 요소와 신흥 기술 추적 및 예측의 차원과 특성을 명확히 제시하였다. 특허 데이터를 기반으로 한 기술 관리, 지식 발견 및 신흥 기술 간의 관계를 분석하여, 신흥 기술을 추적하고 예측하기 위한 특허 데이터 조직 및 지식 발견을 위한 이론적 기반을 구축하고 후속 장의 연구를 위한 이론적 기초를 마련하였다. 제 4 장은 지식 그래프를 활용하여 특허 데이터를 기반으로 한 기술 관리를 구현하며, 이는 제 5, 6, 7장을 지원하여 희토류 산업의 특허 데이터를 통해 실증 분석을 진행하고 특허 데이터를 기술 지식으로 전환하는 역할을 하였다. 제 5, 6, 7 장은 신흥 기술 식별, 진화 및 예측 모델을 구축하고, 이를 희토류 산업의 특허 데이터를 통해 실증적으로 검증하여 신흥 기술의 특징과 패턴을 파악하며, 제 8 장을 위한 기반을 마련하였다. 제 8 장에서는 신흥 기술 추적 및 예측을 위한 특허 데이터를 기반으로 한 기술 관리와 지식 발견을 지원하기 위한 전략과 메커니즘을 제안하며, 국가 기술 혁신을 지원하는 데 중점을 두었다. 주요 연구 기여는 다음과 같다: 1. 신흥 기술을 추적하고 예측하기 위한 특허 데이터를 기반으로 한 기술 관리 및 지식 발견의 메커니즘 모델 구축: 본 연구는 특허와 신흥 기술의 구성 요소 및 특성을 명확히 하여, 어느 특허 데이터 자원이 신흥 기술을 더 잘 나타낼 수 있는지, 그리고 어떤 차원에서 신흥 기술을 추적하고 예측할 수 있는지를 규명한다. 이를 바탕으로 특허와 신흥 기술, 특허 데이터를 기반으로 한 기술 관리와 지식 발견의 관계를 더욱 탐구하며, 데이터 수집, 지식 그래프 기반 데이터 관리 및 지식 발견 단계 간의 상호작용과 원리를 정의한다. 이 모델은 특허 데이터를 기반으로 한 기술 관리 및 지식 발견 과정의 전반적인 개요를 제공하여 후속 실증 연구를 위한 이론적 기초를 형성한다. 2. 특허 지식 그래프 구축: 지식 그래프를 사용하여 기술 데이터를 시각화하고 특허 데이터 관리를 최적화한다. 이를 통해 이전에 분산된 특허 데이터를 복잡한 의미론적 네트워크로 통합하고, 기술 지식의 상호 연결성을 바탕으로 특허 데이터를 저장하고 분석할 수 있다. 이로 인해 특허 데이터의 복잡한 관계를 이해하고 기술 정보를 전체적으로 분석하는 능력이 향상된다. 본 연구는 지식 그래프 기반 데이터 관리의 원칙과 목표를 정의하고, 희토류 산업을 포함한 예시를 통해 특허 지식 그래프를 구축한다. 3. 특허 지식 그래프와 딥러닝을 결합한 신흥 기술 식별 모델 개발: 신흥 기술의 특성과 속성을 철저히 분석한 결과, 신흥 기술 식별 원리를 명확히 하고, 신흥 기술의 새로운 특성, 사회적 영향 및 혁신성을 기반으로 한 정량화된 신흥 기술 지표 시스템을 구축한다. 이를 바탕으로 신흥 기술을 정확히 식별하는 딥러닝 모델을 개발하고, 복합적인 의미론적 정보를 포함한 특허 지식 그래프를 활용하여 식별 과정을 최적화한다. 4. 특허 지식 그래프와 표현 학습을 결합한 신흥 기술 진화 모델 개발: 특허 인용 네트워크, IPC 공존 네트워크 및 기술 의미 정보 등을 바탕으로 신흥 기술 진화 특성을 학습하여 특허 데이터를 기반으로 기술 클러스터를 식별하고, 이를 표현 학습을 통해 미래의 기술 발전 경로를 예측한다. 5. 특허 지식 그래프와 기계 학습을 결합한 신흥 기술 예측 모델 개발: 기술 융합의 내적 메커니즘을 탐구하고, 미래에 발생할 가능성이 있는 신흥 기술 융합을 예측하는 기계 학습 모델을 구축한다. 이를 위해 IPC 공존 네트워크에서 유사성과 융합 지표를 추출하여 KNN, SVM, NB 등 기계 학습 알고리즘을 통해 신흥 기술을 예측한다. 6. 신흥 기술 추적 및 예측을 지원하기 위한 메커니즘 및 전략 제안: 국가 기술 혁신과 주요 기술 돌파구, 경쟁력 확보를 위한 메커니즘과 전략을 수립하여 국가 기술 혁신에 기여하는 특허 데이터 자원 구축, 인재 양성, 기술 혁신 메커니즘을 지원하는 체계적인 지원 시스템을 제안한다. 본 논문은 이론적 측면에서 특허 데이터를 기반으로 한 기술 관리 및 지식 발견의 연구 체계를 확장하고, 실질적인 측면에서 신흥 기술의 식별, 진화, 예측을 통해 국가와 기업의 기술 혁신에 실용적인 지침을 제공한다.

다국어 초록 (Multilingual Abstract)

In the context of global technological innovation, technology management based on patent data and knowledge discovery has become a core strategic necessity for technology development and enhancing competitiveness. The latest international research trends and policy reports further emphasize the impact of emerging technologies on future development. For example, the World Economic Forum’s Top 10 Emerging Technologies Report (2024) highlights that synthetic biology, AI-driven chemistry, and quantum computing will have significant economic and societal impacts in the next five years. Similarly, McKinsey’s Technology Trends Outlook 2024 underscores the importance of generative AI and applied AI in global decarbonization and energy security. These international perspectives reveal the critical position of emerging technologies in global policy and industrial planning, providing valuable references for research on technological innovation. As a crucial part of the technological innovation system, the identification, tracking, prediction, and management of emerging technologies have long been focal points of interest. Accurately tracking and forecasting emerging technologies have become key to driving technological innovation, self-reliance, and national technological independence. Patent data, as the intellectual essence of technological development and innovation, plays a vital role in technological innovation, strategic planning, and emerging technology identification. Thus, technology management based on patent data and knowledge discovery has become both a practical and scientific necessity in the realm of technological innovation. This study delves deeply into emerging technologies through the exploration and analysis of patent data and their interrelationships, focusing on patent elements and emerging technology characteristics. A patent data-driven model for technology management and knowledge discovery is constructed, aimed at tracking and forecasting emerging technologies. By incorporating knowledge graphs, this study leverages their advantages in understanding big data, data relationships, and knowledge organization to systematically analyze and organize patent data, promoting the transformation of patent data into technical knowledge. Utilizing literature analysis, representation learning, machine learning, and empirical research, this study constructs models for technology identification, evolution, and prediction based on patent knowledge graphs, and proposes mechanisms and strategies for technology management and knowledge discovery to track and forecast emerging technologies, supporting national technological innovation strategies. Throughout the research process, Chapter 3 clarifies the components of patent data and the dimensions and characteristics of emerging technology tracking and prediction. By analyzing the relationship between technology management based on patent data, knowledge discovery, and emerging technologies, this chapter builds a theoretical foundation and model for organizing patent data and discovering knowledge for tracking and predicting emerging technologies, providing the theoretical basis for the subsequent chapters. Chapter 4 utilizes knowledge graphs to implement technology management based on patent data, which supports Chapters 5, 6, and 7, facilitating the conversion of patent data into technical knowledge through empirical analysis of patent data in the rare earth industry. Chapters 5, 6, and 7 construct models for the identification, evolution, and prediction of emerging technologies based on patent knowledge graphs and empirically validate these models using patent data from the rare earth industry, identifying the features and patterns of emerging technologies, laying the groundwork for Chapter 8. In Chapter 8, strategies and mechanisms are proposed to support technology management based on patent data and knowledge discovery for tracking and predicting emerging technologies, with a focus on empowering national technological innovation. The primary research contributions are as follows: 1. Construction of a Mechanism Model for Technology Management Based on Patent Data and Knowledge Discovery Aimed at Tracking and Predicting Emerging Technologies: The study clarifies the elements and characteristics of patents and emerging technologies, identifying which patent data resources can better represent emerging technologies, and from which dimensions they can be tracked and predicted. On this basis, the study further explores the relationship between patents, emerging technologies, and technology management based on patent data and knowledge discovery, defining the interaction between different phases—data acquisition, knowledge graph-driven data management, and knowledge discovery—and their respective principles. This model provides a comprehensive overview of the technology management based on patent data and knowledge discovery process, forming the theoretical foundation for subsequent empirical research. 2. Construction of Patent Knowledge Graphs: By utilizing knowledge graphs, technology management based on patent data integrates previously fragmented patent data into a complex semantic network, visualizing and storing patent data based on the interconnections of technical knowledge. This enables a better understanding and analysis of the overall technical information and its interrelations. The study defines the principles and goals of knowledge graph-driven technology management based on patent data and constructs a domain ontology model, incorporating examples from the rare earth industry to create a patent knowledge graph based on Neo4j. Using this graph, patent data is deeply explored, enabling visual patent archives, associated technology analysis, and co-occurrence relationship tracing. 3. Development of an Emerging Technology Identification Model Combining Patent Knowledge Graphs with Deep Neural Networks: Based on a thorough analysis of the features and attributes of emerging technologies, a comprehensive, quantifiable emerging technology index system is developed across three dimensions: novelty, application scope, and development capability. This index system is supported by the complex semantic information in the patent knowledge graph, with feature values extracted through graph query statements and used to train an emerging technology identification model based on deep neural networks and comparative algorithms. 4. Development of an Emerging Technology Evolution Model Combining Patent Knowledge Graphs with Representation Learning: Patent citation networks, IPC co-occurrence networks, and technical semantic information are extracted from the patent knowledge graph, which are then transformed into emerging technology evolution feature vectors using representation learning algorithms. K-means clustering is applied to identify technology groups, and cosine similarity is used to associate these groups across different time periods. The evolution and trends of the identified emerging technologies in the rare earth industry are then analyzed. 5. Development of an Emerging Technology Prediction Model Combining Patent Knowledge Graphs with Machine Learning: The internal mechanism of technological convergence is explored to predict potential future convergence between emerging technology domains. The study identifies the prediction features of emerging technologies from two perspectives: similarity features and fusion indicators. Based on the IPC co-occurrence network, link prediction, association rules, DEMATEL, and semantic correlation methods are used to extract XVIII similarity and fusion features, which are then input into machine learning algorithms to predict future emerging technologies. 6. Proposal of Mechanisms and Strategies to Support Technology Management Based on Patent Data and Knowledge Discovery for Tracking and Predicting Emerging Technologies: From the perspectives of national technological innovation, key technological breakthroughs, and competitive advantage, mechanisms for building patent data resources, cross-disciplinary talent cultivation, and iterative innovation are proposed. These form a four-pronged support system for technology management based on patent data and knowledge discovery, which enhances national technological innovation. This study aims to enrich the theoretical system of technology management based on patent data and knowledge discovery, emphasizing their integration and application in the field of emerging technologies. On a practical level, this study provides scientific and practical reference for identifying emerging technologies, clarifying their evolution, and predicting their future trends, thus offering guidance for technological innovation in national industries and enterprises.

목차 (Table of Contents)

• I. Introduction 1
• 1.1 Research Background 1
• 1.2 Research Significance 5
• 1.3 Overseas and Domestic Research Status 8
• 1.3.1 Technological Innovation Journey of the Rare Earth Industry 9
• 1.3.2 Current Status of Emerging Technology Identification Research 10
• 1.3.3 Current Status of Emerging Technology Evolution Research 12
• 1.3.4 Current Status of Emerging Technology Prediction Research 13
• 1.4 Research Content 18
• 1.4.1 Research Thought 18
• 1.4.2 Main Content 19
• 1.4.3 Research Objectives 22
• 1.5 Research Methods and Research Framework 25
• 1.6 Research Innovations 28
• II. Literature review 30
• 2.1 Relevant Concepts 30
• 2.1.1 Patent-Related Concepts 30
• 2.1.2 Emerging Technologies 31
• 2.1.3 Emerging Technology Tracking and Prediction 33
• 2.2 Theoretical Foundations 36
• 2.2.1 Knowledge Discovery Theory 36
• 2.2.2 Technology Paradigm Theory 37
• 2.2.3 Path Dependence Theory 38
• 2.2.4 Technology Evolution Theory 39
• 2.2.5 Technology Combination Theory 40
• 2.3 Relevant Analysis Methods 42
• 2.3.1 Knowledge Graphs 42
• 2.3.2 Representation Learning 44
• 2.3.3 Link Prediction 46
• 2.3.4 Association Rules 46
• 2.4 Chapter Summary 48
• III. Mechanisms for Emerging Technology Management and Knowledge Discovery 49
• 3.1 Constitutive Elements and Features 50
• 3.1.1 Patent Constitutive Elements 50
• 3.1.2 Constitutive Dimensions of Emerging Technology Tracking and Prediction 53
• 3.1.3 Technical Features of Emerging Technology Tracking and Prediction 56
• 3.2 Relationship Analysis 60
• 3.2.1 Relationship Analysis Between Data Management and Knowledge Discovery 60
• 3.2.2 Analysis of the Relationship Between Patents and Emerging Technologies 61
• 3.3 Process Analysis 65
• 3.3.1 Patent Data Resource Acquisition Phase 65
• 3.3.2 Data Organization Phase Guided by the Patent Knowledge Graph 67
• 3.3.3 Patent Knowledge Discovery Phase Guided by Emerging Technology Tracking and Prediction 68
• 3.4 Mechanism Model of Patent Data Management and Knowledge Discovery 70
• 3.5 Chapter Summary 73
• IV. Construction of a Knowledge Graph for Rare Earth Emerging Technologies 75
• 4.1 Principles and Objectives of Data Management Guided by the Patent Knowledge Graph 76
• 4.1.1 Principles of Data Organization Guided by the Patent Knowledge Graph 76
• 4.1.2 Objectives of Data Organization Guided by the Patent Knowledge Graph 77
• 4.2 Domain Ontology Modeling Based on Patent Data 79
• 4.2.1 Data Sources 79
• 4.2.2 Construction of Core Ontology Classes 82
• 4.2.3 Construction of Patent Domain Ontology Attributes 83
• 4.2.4 Creation of Patent Domain Ontology Instances 88
• 4.3 Storage of the Patent Knowledge Graph Based on Neo4j 92
• 4.3.1 Ontology Mapping and Matching 92
• 4.3.2 Ontology Knowledge Storage 93
• 4.4 Application Scenarios of the Patent Knowledge Graph for Emerging Technology Tracking and Prediction in the Rare Earth Industry 97
• 4.4.1 Construction of Patent Holographic Archives 97
• 4.4.2 Patent Knowledge Retrieval 101
• 4.4.3 Patent Knowledge Graph Inference 103
• 4.5 Chapter Summary 107
• V. Identification of Rare Earth Emerging Technologies 108
• 5.1 Implementation Path of Emerging Technology Identification 110
• 5.1.1 Characteristics of Emerging Technologies 110
• 5.1.2 Attributes and Identification Principles of Emerging Technologies 111
• 5.2 Construction of the Indicator System for Emerging Technology Identification 114
• 5.2.1 Emerging Technology Identification Indicator System 114
• 5.2.2 Calculation of the Emerging Technology Identification Indicator System 115
• 5.3 Construction of the Emerging Technology Identification Model Integrating the Patent Knowledge Graph and Deep Neural Networks 118
• 5.3.1 Feature Extraction Based on the Patent Knowledge Graph 118
• 5.3.2 Classification Labels of Patent Technology Development Potential 120
• 5.3.3 Training the Emerging Technology Identification Model Based on Deep Neural Networks 121
• 5.3.4 Evaluation and Prediction of the Emerging Technology Identification Model 124
• 5.4 Empirical Analysis 126
• 5.4.1 Data Partitioning and Feature Extraction 126
• 5.4.2 Model Training and Evaluation 127
• 5.4.3 Analysis of Identification Results 128
• 5.5 Chapter Summary 134
• VI. Evolution Analysis of Rare Earth Emerging Technologies 137
• 6.1 Analysis of the Evolution Path and Drivers of Emerging Technologies 138
• 6.1.1 Evolution Path of Emerging Technologies 138
• 6.1.2 Analysis of the Drivers of Emerging Technology Evolution 139
• 6.2 Feature Analysis of Emerging Technology Evolution Based on the Patent Knowledge Graph 141
• 6.2.1 Features of Patent Citation Networks 141
• 6.2.2 Features of Patent IPC Co-occurrence Networks 143
• 6.2.3 Semantic Features of Patent Technology 144
• 6.3 Construction of the Emerging Technology Evolution Model Integrating the Patent Knowledge Graph and Representation Learning 145
• 6.3.1 Feature Extraction and Vector Representation 145
• 6.3.2 Identification of Technology Groups Based on K-means Clustering 147
• 6.3.3 Word2vec Embedding Model 148
• 6.3.4 Technology Group Topic Identification 149
• 6.3.5 Analysis of Technology Topic Evolution 150
• 6.4 Empirical Analysis 152
• 6.4.1 Data Selection and Preprocessing 152
• 6.4.2 Evolution Feature Extraction Based on Representation Learning 152
• 6.4.3 Extraction of Technology Groups 154
• 6.4.4 Technology Group Topic Identification 156
• 6.4.5 Analysis of Technology Evolution Path 159
• 6.5 Chapter Summary 168
• VII. Prediction Research on Rare Earth Emerging Technologies 171
• 7.1 The Nature and Mechanism of Emerging Technology Forecasting for Industries 173
• 7.1.1 The Nature of Emerging Technology Forecasting for Industries 173
• 7.1.2 The Mechanism of Emerging Technology Forecasting for Industries 174
• 7.2 Analysis of Features for Predicting Technology Convergence 176
• 7.2.1 Convergence Features Based on Similarity 176
• 7.2.2 Fusion Feature Indicators 178
• 7.3 Construction of the Emerging Technology Prediction Model Integrating the Patent Knowledge Graph and Machine Learning 180
• 7.3.1 Feature Extraction 180
• 7.3.2 Training the Emerging Technology Prediction Model 187
• 7.3.3 Evaluation and Prediction of the Emerging Technology Prediction Model 190
• 7.4 Empirical Analysis 191
• 7.4.1 Data Selection and Preprocessing 191
• 7.4.2 Feature Extraction and Model Training 193
• 7.4.3 Prediction and Analysis of Emerging Technology Prediction Model Results 194
• 7.5 Chapter Summary 203
• VIII. Strategies for Rare Earth Technology Management and Knowledge Discovery 205
• 8.1 Service Scenarios for Technology Management and Knowledge Discovery 206
• 8.1.1 Empowering National Technological Innovation 206
• 8.1.2 Supporting Breakthroughs in Key Technologies 210
• 8.1.3 Supporting the Seizing of Competitive Advantages 211
• 8.2 Protection Mechanisms for Technology Management and Knowledge Discovery 214
• 8.2.1 Mechanism for Building Patent Data Resources 211
• 8.2.2 Cross-disciplinary Talent Support Mechanism 215
• 8.2.3 Iterative Embedding Mechanism for Intelligent Technologies 216
• 8.2.4 Mechanism for Strengthening Technology Innovation Platforms 218
• 8.3 Strategies for Technology Management and Knowledge Discover 220
• 8.3.1 Safeguard Strategies from a Perspective of Collaborative Action 221
• 8.3.2 Strategies from the Perspective of Technological Development 224
• 8.3.3 Strategies from the Perspective of Multi-stakeholders 225
• 8.4 Chapter Summary 228
• IX. Conclusions 229
• 9.1 Key Research Findings 230
• 9.2 Research Innovations 232
• 9.3 Research Limitations 234
• 9.4 Research Outlook 237
• REFERENCES 239
• APPENDIX 258
• ACKNOWLEDGEMENTS 261