Font Size: a A A

Research Of Disciplinary Metrology Methods Based On Representation Learning

Posted on:2020-07-13Degree:MasterType:Thesis
Country:ChinaCandidate:H F LiuFull Text:PDF
GTID:2439330590496933Subject:Science and technology management
Abstract/Summary:PDF Full Text Request
Discipline is the basic structure level of the scientific knowledge system,academic journals are the main media of scientific activities and one of the main objects for scientometrists to study the laws of science.As the main carriers of scientific knowledge,academic papers contain the research results of scholars and the laws of scientific cooperation in academic papers.Quantify the representation of disciplines,mining potential related information of journals and topics in disciplines,and accurately calculated expressing information such as disciplines,journals,and topics through numerical vectors,and further exploring the implications of disciplines,underlying scientific laws.Analyzing the internal structure of the scientific field and establishing a comprehensive and scientific knowledge research model is not only important for understanding the science theoretical system,but also have great practical significance for grasping scientific development trend,formulate scientific research projects and technology policies.Based on the representational learning model,this paper proposes the disciplinary representation method,such as the discipline to discipline method,the discipline to journals method,and the discipline to topics method.The disciplines,journals,topics,and other information are integrated to construct a discipline measurement method suitable for the scientific metrology research system.By representing the vector to capture the information entity containing the inherent laws of science,the paper analyzes the interdisciplinary analysis,the definition of the subject core journal,and the potential cooperation opportunities of scholars,and demonstrates the feasibility of representing the learning model in the scientific field.Based on the proposed method of measuring and representing the disciplines,the author analyzes the characteristics of the multidisciplinary development of the current disciplines.The disciplinary system can be divided into eight clusters of disciplines.For different disciplines,the journals can be used to identify the core journals in the disciplines.At the same time,the knowledge representation learning model is used at the discipline to conduct core scholar identification and potential cooperation opportunities.The discipline measurement method proposed in this paper can provide constructive opinions for the construction of the discipline system and the formulation of national science and technology policies,as well as the construction of talent think tanks.
Keywords/Search Tags:Scientometrics, Categories Representation, Disciplinary metrology, Network representation learning, Knowledge representation learning
PDF Full Text Request
Related items