COMMON NEIGHBORHOOD-BASED LINK PREDICTION IN SPORTS NETWORKS

Emrah Özkaynak

Makale detayları

COMMON NEIGHBORHOOD-BASED LINK PREDICTION IN SPORTS NETWORKS

Yazarlar:

Emrah Özkaynak

Yayıncı:

Karabuk University

Ana Başlıklar:

Pattern Recognition | Modelling and Simulation

Dergi:

Current Trends in Computing Vol.1 Issue 1, pp.10-21

Belge Tipi:

Makale

Yayın Tarihi:

2/2024

Belge Dili:

İngilizce

Document ID (Org No.):

2001311310

pISSN:

2980-3152

Anahtar kelimeler:

Özet:

 Link prediction has been among the popular topics in social network analysis studies in recent years. The prediction of new links that may arise in the future, depending on the analysis of the relations between the entities, has started to be used frequently, especially in recommendation systems. Link prediction methods, especially used in social networks, mostly use the topological features of complex networks in terms of application. This situation has also paved the way for link prediction methods to be preferred in almost all kinds of networks of complex network structures. The increased trend in link prediction studies has also allowed many methods to be proposed and used in this field. The differences in the formation of the network and the link types prevent the developed methods from giving the same performance for every complex network. This situation has increased the importance of choosing the appropriate link prediction method depending on the structure of the complex network. This study applied neighborhood-based link prediction methods in networks created from different sports competitions. Furthermore, The most suitable neighborhood-based link prediction method that could be used in sports networks has been investigated. Link prediction methods were applied to the networks formed with different time periods formed from different sports branches such as tennis tournaments, football competitions, and billiards competitions, and the accuracy performances of the methods were determined. The results obtained from the AUC metric in the experimental studies show that the neighborhood-based link prediction methods successfully predict the new connections that may arise in the future in sports networks.

 Link prediction has been among the popular topics in social network analysis studies in recent years. The prediction of new links that may arise in the future, depending on the analysis of the relations between the entities, has started to be used frequently, especially in recommendation systems. Link prediction methods, especially used in social networks, mostly use the topological features of complex networks in terms of application. This situation has also paved the way for link prediction methods to be preferred in almost all kinds of networks of complex network structures. The increased trend in link prediction studies has also allowed many methods to be proposed and used in this field. The differences in the formation of the network and the link types prevent the developed methods from giving the same performance for every complex network. This situation has increased the importance of choosing the appropriate link prediction method depending on the structure of the complex network. This study applied neighborhood-based link prediction methods in networks created from different sports competitions. Furthermore, The most suitable neighborhood-based link prediction method that could be used in sports networks has been investigated. Link prediction methods were applied to the networks formed with different time periods formed from different sports branches such as tennis tournaments, football competitions, and billiards competitions, and the accuracy performances of the methods were determined. The results obtained from the AUC metric in the experimental studies show that the neighborhood-based link prediction methods successfully predict the new connections that may arise in the future in sports networks.

 Link prediction has been among the popular topics in social network analysis studies in recent years. The prediction of new links that may arise in the future, depending on the analysis of the relations between the entities, has started to be used frequently, especially in recommendation systems. Link prediction methods, especially used in social networks, mostly use the topological features of complex networks in terms of application. This situation has also paved the way for link prediction methods to be preferred in almost all kinds of networks of complex network structures. The increased trend in link prediction studies has also allowed many methods to be proposed and used in this field. The differences in the formation of the network and the link types prevent the developed methods from giving the same performance for every complex network. This situation has increased the importance of choosing the appropriate link prediction method depending on the structure of the complex network. This study applied neighborhood-based link prediction methods in networks created from different sports competitions. Furthermore, The most suitable neighborhood-based link prediction method that could be used in sports networks has been investigated. Link prediction methods were applied to the networks formed with different time periods formed from different sports branches such as tennis tournaments, football competitions, and billiards competitions, and the accuracy performances of the methods were determined. The results obtained from the AUC metric in the experimental studies show that the neighborhood-based link prediction methods successfully predict the new connections that may arise in the future in sports networks.

 Link prediction has been among the popular topics in social network analysis studies in recent years. The prediction of new links that may arise in the future, depending on the analysis of the relations between the entities, has started to be used frequently, especially in recommendation systems. Link prediction methods, especially used in social networks, mostly use the topological features of complex networks in terms of application. This situation has also paved the way for link prediction methods to be preferred in almost all kinds of networks of complex network structures. The increased trend in link prediction studies has also allowed many methods to be proposed and used in this field. The differences in the formation of the network and the link types prevent the developed methods from giving the same performance for every complex network. This situation has increased the importance of choosing the appropriate link prediction method depending on the structure of the complex network. This study applied neighborhood-based link prediction methods in networks created from different sports competitions. Furthermore, The most suitable neighborhood-based link prediction method that could be used in sports networks has been investigated. Link prediction methods were applied to the networks formed with different time periods formed from different sports branches such as tennis tournaments, football competitions, and billiards competitions, and the accuracy performances of the methods were determined. The results obtained from the AUC metric in the experimental studies show that the neighborhood-based link prediction methods successfully predict the new connections that may arise in the future in sports networks.

Makale detayları

COMMON NEIGHBORHOOD-BASED LINK PREDICTION IN SPORTS NETWORKS

Yazarlar:

Emrah Özkaynak

Yayıncı:

Karabuk University

Ana Başlıklar:

Pattern Recognition | Modelling and Simulation

Dergi:

Current Trends in Computing Vol.1 Issue 1, pp.10-21

Belge Tipi:

Makale

Yayın Tarihi:

2/2024

Belge Dili:

İngilizce

Document ID (Org No.):

2001311310

pISSN:

2980-3152

Link Prediction

Social Networks

Data Mining.

Sayı

Makale detayları

COMMON NEIGHBORHOOD-BASED LINK PREDICTION IN SPORTS NETWORKS

Yazarlar: Emrah Özkaynak

Yayıncı: Karabuk University

Ana Başlıklar: Pattern Recognition | Modelling and Simulation

Dergi: Current Trends in Computing Vol.1 Issue 1, pp.10-21

Belge Tipi: Makale

Yayın Tarihi: 2/2024

Belge Dili: İngilizce

Document ID (Org No.): 2001311310

pISSN: 2980-3152

Link Prediction

Social Networks

Data Mining.

Sayı

Yazarlar:

Emrah Özkaynak

Yayıncı:

Karabuk University

Ana Başlıklar:

Pattern Recognition | Modelling and Simulation

Dergi:

Current Trends in Computing Vol.1 Issue 1, pp.10-21

Belge Tipi:

Makale

Yayın Tarihi:

2/2024

Belge Dili:

İngilizce

Document ID (Org No.):

2001311310

pISSN:

2980-3152