Event detection in scientific mapping based on a novel structural community similarity algorithm

Meng Xiangfeng, Liu Xinhai, Zhang Yan, W. Glänzel

Research output: Contribution to conferencePaper

Abstract

Science is a dynamic system. Its cognitive structure is subject to permanent changes. Any particular structural change, such as new research trends and emerging topics, can be detected in cognitive maps. In this paper, we attempt to identify new research trends and topic transition by detecting both changing structure and anomalous events in the course of mapping the structure of journal-based clustering. We form one community for each journal cluster, with nodes representing journals and edges representing the citation links among journals. To measure the significance of changes in the dynamic journal clusters, a community similarity measurement is required. However, the existing similarity algorithms neglect the shift of the internal structure in networks and graphs, which determines that these algorithms cannot be directly applied to journal clusters. We propose a novel community-similarity algorithm, which considers both the shifts of vertices and the shifts of communities' layered structure. Communities' layered structure categorizes nodes into different groups, depending on their influence on the community. We apply the novel algorithm on the temporal journal data set, and identify two types of anomalous events. Both the visualizations of the journal clusters and the text annotations demonstrate that the identified events correspond to new emerging trends in scientific fields.

Original languageEnglish
Pages258-269
Number of pages12
Publication statusPublished - Jan 1 2017
Event16th International Conference on Scientometrics and Informetrics, ISSI 2017 - Wuhan, China
Duration: Oct 16 2017Oct 20 2017

Other

Other16th International Conference on Scientometrics and Informetrics, ISSI 2017
CountryChina
CityWuhan
Period10/16/1710/20/17

Fingerprint

Event Detection
Anomalous
Cognitive Map
Dynamical systems
Visualization
Citations
Structural Change
Vertex of a graph
Dynamic Systems
Annotation
Community
Similarity
Event detection
Clustering
Internal
Graph in graph theory
Demonstrate
Trends
Node
Community structure

Keywords

  • Knowledge discovery and data mining
  • Scientometrics

ASJC Scopus subject areas

  • Computer Science Applications
  • Applied Mathematics
  • Modelling and Simulation
  • Statistics and Probability
  • Management Science and Operations Research

Cite this

Xiangfeng, M., Xinhai, L., Yan, Z., & Glänzel, W. (2017). Event detection in scientific mapping based on a novel structural community similarity algorithm. 258-269. Paper presented at 16th International Conference on Scientometrics and Informetrics, ISSI 2017, Wuhan, China.

Event detection in scientific mapping based on a novel structural community similarity algorithm. / Xiangfeng, Meng; Xinhai, Liu; Yan, Zhang; Glänzel, W.

2017. 258-269 Paper presented at 16th International Conference on Scientometrics and Informetrics, ISSI 2017, Wuhan, China.

Research output: Contribution to conferencePaper

Xiangfeng, M, Xinhai, L, Yan, Z & Glänzel, W 2017, 'Event detection in scientific mapping based on a novel structural community similarity algorithm' Paper presented at 16th International Conference on Scientometrics and Informetrics, ISSI 2017, Wuhan, China, 10/16/17 - 10/20/17, pp. 258-269.
Xiangfeng M, Xinhai L, Yan Z, Glänzel W. Event detection in scientific mapping based on a novel structural community similarity algorithm. 2017. Paper presented at 16th International Conference on Scientometrics and Informetrics, ISSI 2017, Wuhan, China.
Xiangfeng, Meng ; Xinhai, Liu ; Yan, Zhang ; Glänzel, W. / Event detection in scientific mapping based on a novel structural community similarity algorithm. Paper presented at 16th International Conference on Scientometrics and Informetrics, ISSI 2017, Wuhan, China.12 p.
@conference{d385707c73a94a0db01925472479f0ab,
title = "Event detection in scientific mapping based on a novel structural community similarity algorithm",
abstract = "Science is a dynamic system. Its cognitive structure is subject to permanent changes. Any particular structural change, such as new research trends and emerging topics, can be detected in cognitive maps. In this paper, we attempt to identify new research trends and topic transition by detecting both changing structure and anomalous events in the course of mapping the structure of journal-based clustering. We form one community for each journal cluster, with nodes representing journals and edges representing the citation links among journals. To measure the significance of changes in the dynamic journal clusters, a community similarity measurement is required. However, the existing similarity algorithms neglect the shift of the internal structure in networks and graphs, which determines that these algorithms cannot be directly applied to journal clusters. We propose a novel community-similarity algorithm, which considers both the shifts of vertices and the shifts of communities' layered structure. Communities' layered structure categorizes nodes into different groups, depending on their influence on the community. We apply the novel algorithm on the temporal journal data set, and identify two types of anomalous events. Both the visualizations of the journal clusters and the text annotations demonstrate that the identified events correspond to new emerging trends in scientific fields.",
keywords = "Knowledge discovery and data mining, Scientometrics",
author = "Meng Xiangfeng and Liu Xinhai and Zhang Yan and W. Gl{\"a}nzel",
year = "2017",
month = "1",
day = "1",
language = "English",
pages = "258--269",
note = "16th International Conference on Scientometrics and Informetrics, ISSI 2017 ; Conference date: 16-10-2017 Through 20-10-2017",

}

TY - CONF

T1 - Event detection in scientific mapping based on a novel structural community similarity algorithm

AU - Xiangfeng, Meng

AU - Xinhai, Liu

AU - Yan, Zhang

AU - Glänzel, W.

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Science is a dynamic system. Its cognitive structure is subject to permanent changes. Any particular structural change, such as new research trends and emerging topics, can be detected in cognitive maps. In this paper, we attempt to identify new research trends and topic transition by detecting both changing structure and anomalous events in the course of mapping the structure of journal-based clustering. We form one community for each journal cluster, with nodes representing journals and edges representing the citation links among journals. To measure the significance of changes in the dynamic journal clusters, a community similarity measurement is required. However, the existing similarity algorithms neglect the shift of the internal structure in networks and graphs, which determines that these algorithms cannot be directly applied to journal clusters. We propose a novel community-similarity algorithm, which considers both the shifts of vertices and the shifts of communities' layered structure. Communities' layered structure categorizes nodes into different groups, depending on their influence on the community. We apply the novel algorithm on the temporal journal data set, and identify two types of anomalous events. Both the visualizations of the journal clusters and the text annotations demonstrate that the identified events correspond to new emerging trends in scientific fields.

AB - Science is a dynamic system. Its cognitive structure is subject to permanent changes. Any particular structural change, such as new research trends and emerging topics, can be detected in cognitive maps. In this paper, we attempt to identify new research trends and topic transition by detecting both changing structure and anomalous events in the course of mapping the structure of journal-based clustering. We form one community for each journal cluster, with nodes representing journals and edges representing the citation links among journals. To measure the significance of changes in the dynamic journal clusters, a community similarity measurement is required. However, the existing similarity algorithms neglect the shift of the internal structure in networks and graphs, which determines that these algorithms cannot be directly applied to journal clusters. We propose a novel community-similarity algorithm, which considers both the shifts of vertices and the shifts of communities' layered structure. Communities' layered structure categorizes nodes into different groups, depending on their influence on the community. We apply the novel algorithm on the temporal journal data set, and identify two types of anomalous events. Both the visualizations of the journal clusters and the text annotations demonstrate that the identified events correspond to new emerging trends in scientific fields.

KW - Knowledge discovery and data mining

KW - Scientometrics

UR - http://www.scopus.com/inward/record.url?scp=85036614790&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85036614790&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:85036614790

SP - 258

EP - 269

ER -