scholarly journals A Method for Improving the Accuracy of Link Prediction Algorithms

Complexity ◽  
2021 ◽  
Vol 2021 ◽  
pp. 1-5
Author(s):  
Jie Li ◽  
Xiyang Peng ◽  
Jian Wang ◽  
Na Zhao
Keyword(s):  
Social Networks ◽  
Complex Networks ◽  
Random Walks ◽  
Link Prediction ◽  
Prediction Accuracy ◽  
Decomposition Method ◽  
Local Information ◽  
Evolution Mechanism ◽  
Real Network ◽  
Prediction Algorithms

Link prediction is a key tool for studying the structure and evolution mechanism of complex networks. Recommending new friend relationships through accurate link prediction is one of the important factors in the evolution, development, and popularization of social networks. At present, scholars have proposed many link prediction algorithms based on the similarity of local information and random walks. These algorithms help identify actual missing and false links in various networks. However, the prediction results significantly differ in networks with various structures, and the prediction accuracy is low. This study proposes a method for improving the accuracy of link prediction. Before link prediction, k-shell decomposition method is used to layer the network, and the nodes that are in 1-shell and the nodes that are not linked to the high-shell in the 2-shell are deleted. The experiments on four real network datasets verify the effectiveness of the proposed method.

ZERO NODES EFFECT: VALID LINK PREDICTION IN SPARSE NETWORKS

2013 ◽  
Vol 27 (12) ◽  
pp. 1350052 ◽  
Author(s):  
CHEN-XI SHAO ◽  
HUI-LING DOU ◽  
RONG-XU YANG ◽  
BING-HONG WANG
Keyword(s):  
Complex Networks ◽  
Link Prediction ◽  
Prediction Accuracy ◽  
Preferential Attachment ◽  
Hybrid Index ◽  
Evolution Mechanism ◽  
Evolutionary Forces ◽  
Similarity Indices ◽  
Zero Degree ◽  
Sparse Networks

Zero-degree nodes are an important difficulty in sparse networks' link prediction. Clustering and preferential attachment, as the most important characteristics of complex networks, have been paid little attention in similarity indices. Inspired by the coexistence of clustering and preferential attachment in real networks, this paper proposes a new preferential attachment index and new clustering index, which have here been integrated into a hybrid index that considers the dynamic evolutionary forces of complex networks and can solve the problem of excessive zero-degree nodes in sparse networks and check evolution mechanism. Experiments proved prediction accuracy can be remarkably enhanced.

scholarly journals A novel similarity measure for missing link prediction in social networks

2020 ◽  
Vol 19 (2) ◽  
pp. 1071
Author(s):  
Gogulamudi Naga Chandrika ◽  
E. Srinivasa Reddy
Keyword(s):  
Social Networks ◽  
Complex Networks ◽  
Similarity Measure ◽  
Network Theory ◽  
Link Prediction ◽  
Prediction Methods ◽  
Missing Link ◽  
The Social ◽  
Hidden Patterns ◽  
Over Time

<p><span>Social Networks progress over time by the addition of new nodes and links, form associations with one community to the other community. Over a few decades, the fast expansion of Social Networks has attracted many researchers to pay more attention towards complex networks, the collection of social data, understand the social behaviors of complex networks and predict future conflicts. Thus, Link prediction is imperative to do research with social networks and network theory. The objective of this research is to find the hidden patterns and uncovered missing links over complex networks. Here, we developed a new similarity measure to predict missing links over social networks. The new method is computed on common neighbors with node-to-node distance to get better accuracy of missing link prediction. </span><span>We tested the proposed measure on a variety of real-world linked datasets which are formed from various linked social networks. The proposed approach performance is compared with contemporary link prediction methods. Our measure makes very effective and intuitive in predicting disappeared links in linked social networks.</span></p>

Similarity indices based on link weight assignment for link prediction of unweighted complex networks

2017 ◽  
Vol 31 (02) ◽  
pp. 1650254 ◽  
Author(s):  
Shuxin Liu ◽  
Xinsheng Ji ◽  
Caixia Liu ◽  
Yi Bai
Keyword(s):  
Complex Networks ◽  
Link Prediction ◽  
Prediction Accuracy ◽  
Prediction Methods ◽  
Close Attention ◽  
Common Neighbor ◽  
Link Weight ◽  
Similarity Indices ◽  
Clustered Networks ◽  
Local Path

Many link prediction methods have been proposed for predicting the likelihood that a link exists between two nodes in complex networks. Among these methods, similarity indices are receiving close attention. Most similarity-based methods assume that the contribution of links with different topological structures is the same in the similarity calculations. This paper proposes a local weighted method, which weights the strength of connection between each pair of nodes. Based on the local weighted method, six local weighted similarity indices extended from unweighted similarity indices (including Common Neighbor (CN), Adamic-Adar (AA), Resource Allocation (RA), Salton, Jaccard and Local Path (LP) index) are proposed. Empirical study has shown that the local weighted method can significantly improve the prediction accuracy of these unweighted similarity indices and that in sparse and weakly clustered networks, the indices perform even better.

Prediction of missing links and reconstruction of complex networks

2016 ◽  
Vol 27 (10) ◽  
pp. 1650120 ◽  
Author(s):  
Cheng-Jun Zhang ◽  
An Zeng
Keyword(s):  
Complex Networks ◽  
Link Prediction ◽  
Prediction Accuracy ◽  
High Performance ◽  
Point Of View ◽  
Prediction Methods ◽  
Online Systems ◽  
Network Properties ◽  
High Prediction ◽  
Simple Link

Predicting missing links in complex networks is of great significance from both theoretical and practical point of view, which not only helps us understand the evolution of real systems but also relates to many applications in social, biological and online systems. In this paper, we study the features of different simple link prediction methods, revealing that they may lead to the distortion of networks’ structural and dynamical properties. Moreover, we find that high prediction accuracy is not definitely corresponding to a high performance in preserving the network properties when using link prediction methods to reconstruct networks. Our work highlights the importance of considering the feedback effect of the link prediction methods on network properties when designing the algorithms.

scholarly journals Link prediction based on the powerful combination of endpoints and neighbors

2020 ◽  
Vol 34 (28) ◽  
pp. 2050269
Author(s):  
Tianrun Gao ◽  
Xuzhen Zhu
Keyword(s):  
Complex Networks ◽  
Performance Improvement ◽  
Link Prediction ◽  
Prediction Models ◽  
Important Research ◽  
Topological Similarity ◽  
Real Network ◽  
Model Experiments ◽  
Large Influence ◽  
Proposed Model

Performance improvement of topological similarity-based link prediction models becomes an important research in complex networks. In the models based on node influence, researchers mainly consider the roles of endpoints or neighbors. Through investigations, we find that an endpoint with large influence has many neighbors. Meanwhile, the neighbors connect with more nodes besides endpoint, meaning that the endpoint can transmit extensive influence by the powerful combination of itself and neighbors. In addition, we evaluate the node influence by degree because the degree represents the number of neighbors accurately. In this paper, through focusing on the degree of endpoints and neighbors, we propose the powerful combination of endpoints and neighbors (PCEN) model. Experiments on twelve real network datasets demonstrate that the proposed model has better prediction performances than the traditional models.

Cyber Buzz: Examining Virality Characteristics of Cybersecurity Content In Social Networks

2020 ◽  
Vol 64 (1) ◽  
pp. 441-445
Author(s):  
Thomas Shields ◽  
Hannah Li ◽  
Peter Lebedev ◽  
Josiah Dykstra
Keyword(s):  
Social Networks ◽  
Social Network ◽  
Online Social Networks ◽  
Prediction Accuracy ◽  
Time Of Day ◽  
Policy Adoption ◽  
The Internet ◽  
Security Awareness ◽  
Prediction Algorithms

The Internet is a rich environment for information to spread rapidly and widely. The ability of cybersecurity content to achieve virality in social networks can be useful for measuring security awareness, policy adoption, or cybersecurity literacy. It may also reveal new and emerging cybersecurity events. Virality in online social networks can be characterized and measured many ways and have different causes. Leveraging existing research in social network virality measurements, we calculate and analyze virality measurements and correlations on an anonymized Reddit dataset, examining overall trends and characteristics of individual cybersecurity forums (subreddits). We reproduce content-based virality prediction algorithms and assess their performance, then introduce additional features beyond post title, including time of day, to improve prediction accuracy to ~71% for each of the virality scores. We examine the intersection of the virality facets to reveal correlations about the content and times when cybersecurity content is most viral.

Link prediction in complex networks using information-theoretic measures

2020 ◽  
Vol 8 (4) ◽  
Author(s):  
Deepanshu Malhotra ◽  
Rinkaj Goyal
Keyword(s):  
Mutual Information ◽  
Complex Networks ◽  
Link Prediction ◽  
Structural Features ◽  
Engineering Systems ◽  
Form Complex ◽  
Information Theoretic ◽  
Information Measures ◽  
Prediction Algorithms ◽  
Information Theoretic Measures

Abstract Interconnections among real-world entities through explicit or implicit relationships form complex networks, such as social, economic and engineering systems. Recently, the studies based on such complex networks have provided a boost to our understanding of various events and processes ranging from biology to technology. Link prediction algorithms assist in predicting, analysing and deciphering more significant details about the networks and their future structures. In this study, we propose three different link prediction algorithms based on different structural features of the network combined with the information-theoretic analyses. The first two algorithms (variants) are developed for unweighted networks, while the third approach deals with the weighted ones. The proposed methods exhibit better and robust performances in the majority of cases, and at least comparable, if not better in other cases. This work is built upon the previously published mutual information-based approaches for link prediction; however, this study considers structural features of the network to augment mutual information measures and provides insights for finding hidden links in the network.

scholarly journals Link Prediction Algorithms for Social Networks Based on Machine Learning and HARP

IEEE Access ◽  
2019 ◽  
Vol 7 ◽  
pp. 122722-122729 ◽  
Author(s):  
Hao Shao ◽  
Lunwen Wang ◽  
Yufan Ji
Keyword(s):  
Machine Learning ◽  
Social Networks ◽  
Link Prediction ◽  
Prediction Algorithms
Sign in / Sign up

Export Citation Format

Share Document