Multi-source data fusion study in scientometrics

Abstract

This paper provides an introduction to multi-source data fusion (MSDF) and comprehensively overviews the ingredients and challenges of MSDF in scientometrics. As compared to the MSDF methods in the sensor and other fields, and considering the features of scientometrics, in this paper an application model and procedure of MSDF in scientometrics are proposed. The model and procedure can be divided into three parts: data type integration, fusion of data relations, and ensemble clustering. Furthermore, the fusion of data relations can be divided into cross-integration of multi-mode data and matrix fusion of multi-relational data. To obtain a clearer and deeper analysis of the MSDF model, this paper further focuses on the application of MSDF in topic identification based on text analysis of scientific literatures. This paper also discusses the application of MSDF for the exploration of scientific literatures. Finally, the most suitable MSDF methods for different situations are discussed.

Analysis of Hot Topics and Evolution of Research in World-class Agricultural Universities Based on BERTopic

“…[Relationship between science and technology] The connection and significance of cited papers and sizing patents Papers are often used to measure the level of research activity in the basic sciences, while patents are often used to measure the level of innovation in industrial technologies [6][7][8]. Increasingly, patents are generating innovations that advance technological progress and contribute to economic growth by directly citing papers [9], and the notion that technology stems from a scientific base was initially embedded in linear or pipeline models.…”

Research on the Construction Path of Vocational Education Virtual Simulation Training Base Combining Multi-source Data

“…There are many ways of multi-source data fusion, either from the underlying semantic framework, or from the data structure perspective, or from the number theory perspective [17]. This paper analyzes them from the perspective of data structure.…”

Data-Driven Computational Social Science: A Survey

“…Liu et al [173] propose a general multi-source data fusion framework that solves the heterogeneity and semantic conflict in the big data environment. In internet of things [174], [175], scientometrics [176], network science [177] and material science areas [178], researchers have designed diverse multi-source data fusion methods which aim at their own discipline characteristics and issues. In the future, the multi-source data fusion problem will still be challenging under the circumstances of the rapid growth of data and the interactions between offline and online.…”