Negative sign prediction for signed social networks
Introduction
Social networks in many real world applications are signed social networks [1], i.e., links between nodes of these social networks have positive or negative signs. Positive sign means two end nodes of the link are positively related [2], e.g., they trust each other, they are friends or they agree with each other. Negative sign means two end nodes of the link are negatively related [3], e.g., they distrust each other, they are foes, or they disagree with other each other.
Sign prediction predicts signs for links of signed social network. It reveals the underlying relationships between users. It can be widely used in many applications such as recommendation systems and abnormal user detections [4]. Most works of sign prediction focus on positive sign prediction [5], [6]. The research of negative sign prediction is still at the beginning stage. This is because: (a) It is easier to predict positive signs than predict negative signs [7]. Positive signs can be propagated between users of social networks. E.g., if A trusts B and B trusts C, A will trust C to some extent. While negative signs cannot be propagated between users of social networks, which makes negative sign prediction more complex. E.g., if A distrusts B and B distrusts C, it is hard to judge the relationships between A and C directly. (b) The number of publicly available datasets for negative sign prediction research is limited [8]. Compared with positive signs, the information of negative signs between users of social networks is more sensitive. So there are only several available real world datasets containing negative signs.
Though the research of negative sign prediction is far less than the research of positive sign prediction, it is extremely important in the real world applications. The effects of negative signs and the effects of positive signs are unbalanced in signed social networks [9], [10]. In real world applications, the number of negative signs is far less than the number of positive signs. Users rarely express their antipathy to others due to their manners or fear of being retaliated [11]. However, it is the negative signs which dominate the final decision made by users in most cases. E.g., in the recommendation system, even if 100 users express positive opinions on a target user, a user may not choose the items recommended by this user if several users express negative opinions on this target user. Existing works [11], [12], [13], [14], [15] of negative sign prediction are based on Structural Balance Theory [16], Social Status Theory [17], or both. Structural Balance Theory and Social Status Theory analyze the triad relationship between nodes of networks, which include positive relationships and negative relationships. Structural Balance Theory regards the enemy of enemy as friend, and regards the enemy of friend as enemy [16]. It considers four possible triad relationships with positive and/or negative relationships in signed networks. Two of these triad relationships are regarded as balanced. Social Status Theory regards the sign between users as the gap between their statuses [17]. A positive sign for the link pointing from A to B means A thinks the status of B is higher than A, while a negative sign means A thinks the status of B is lower than A. Social Status Theory therefore defines eight balanced triad relationships in signed networks. Existing works use some attributes based on Structural Balance Theory and/or Social Status Theory and predict the negative signs. However, the attributes evolved in existing works only reflect partial information related to negative signs, which leads to the limited negative sign prediction accuracy. Attributes reflect different aspects of signs shall be considered comprehensively to get better prediction performances in the negative sign predictions.
To solve the problems of existing works, we propose a novel negative sign prediction method which involves negative sign related features of different categories. The proposed method contributes to generalize three main categories of negative sign related features: nodes features which reflect a node’s tendency of giving/receiving negative signs, triad features which reflect the properties of nodes according to Structural Balance Theory and Social Status Theory, and user similarity features which reflect the similarities between users giving/receiving negative signs to/from the target user. The proposed model totally selects 7 features from these three categories: negative indegree ratio, negative outdegree ratio, negative triad ratio, source node positive similarity, source node negative similarity, destination node positive similarity and destination node negative similarity. Logistic Regression Model is used to merge these 7 features to predict negative signs for signed social networks. Experiments held on three publicly available datasets verified the effectiveness of the proposed method: compared with existing works, the negative sign prediction accuracy can be improved around , while the negative sign prediction F1 score can be significantly improved, up to . The generalization performance of the proposed method is also superior to existing works, up to improvement comparing with existing works. In addition, the proposed method also has better prediction accuracy and F1 score for signed social networks with different embeddedness, especially for signed social networks whose embeddedness equals to 0.
The rest of the paper is organized as follows: Section 2 introduces the related works, Section 3 presents the proposed negative sign prediction method in details, Section 4 gives the experimental results of the proposed method, and Section 5 concludes this paper and points out the future directions.
Section snippets
Related works
In the field of sign prediction, there are two fundamental theories. Existing works adopted one of them, both of them or extended these theories. These two fundamental theories are:
(1) Structural Balance Theory
In Structural Balance Theory [16], there are four distinct ways to label the three edges among three nodes with positive signs and negative signs. Given a set of nodes and .
A. Having three positive signs, as shown in Fig. 1 (a). This means these nodes are mutual friends.
B. Having a
The proposed method
Let a graph represent the signed social network, where is the set of nodes representing users of signed social network, and is the set of edges representing relationships between users of signed social network. represents the edge pointing from node to node , . does not necessarily equal to , i.e., edges in the signed social networks have directions and is a directed graph. has three possible values, which is defined as follows:
Experimental results
Experiments are held on three real world datasets to verify the performance of the proposed method. The datasets are extracted from three online social networks Epinions, Slashdot and Wikipedia respectively. The datasets are publicly available at [20]. The detailed information of these datasets is given in Table 1. Epinions dataset has 131 828 nodes and 841 372 edges, in which 123 705 edges have negative signs. Slashdot dataset has 82 140 nodes and 549,202 edges, in which 124 130 edges have
Conclusion and future works
Negative sign prediction predicts the existence of negative signs for the links of signed social networks. Though the number of negative signs is far less than the number of positive signs in signed social networks, negative signs are much more important for the user decision. Negative signs cannot be propagated between users like positive signs. So positive sign prediction models cannot be used in negative sign prediction. Negative signs are usually predicted according to Structural Balance
Acknowledgments
This research was supported by Nature Science Foundation of China (Grant No. 61672284). This work was also supported by China Postdoctoral Science Foundation (Grant No. 2016M591841), Open fund of State Key Laboratory of Acoustics (no. SKLA201706), and Open Project Foundation of Information Technology Research Base of Civil Aviation Administration of China (Grant No. CAAC-ITRB-201501 and Grant No. CAAC-ITRB-201602).
Weiwei Yuan is an associate professor in College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, China. She received her B.S. and M.S. degree from Harbin Engineering University, China in 2002 and 2005 respectively. She received her Ph.D. degree in Department of Computer Engineering, Kyung Hee University, South Korea in 2010. Her research interests include Pattern Recognition, Social Computing and Recommender Systems.
References (27)
- et al.
Partition signed social networks via clustering dynamics
Physica A
(2016) - et al.
The influence of structural balance and homophily/heterophobia on the adjustment of random complete signed networks
Social Networks
(2016) - et al.
Improved trust-aware recommender system using small-worldness of trust networks
Knowl.-Based Syst.
(2010) - et al.
LED: A fast overlapping communities detection algorithm based on structural clustering
Neurocomputing
(2016) - J. Tang, C. Aggarwal, H. Liu, Recommendations in signed social networks, in: International conference on world wide...
- et al.
The evolution of beliefs over signed social networks
Oper. Res.
(2013) - J. Tang, C. Aggarwal, H. Liu, Node classification in signed social networks, in: SIAM International Conference on Data...
- et al.
Unsupervised Sentiment Analysis with Signed Social Networks
- et al.
Maximizing the spread of positive influence in signed social networks
Intell. Data Anal.
(2016) - et al.
A survey of signed network mining in social media
ACM Comput. Surv.
(2016)
Model-Based Collaborative Personalized Recommendation on Signed Social Rating Networks
Cited by (13)
Uncovering the trade-induced air pollution inequality among Belt and Road Initiative (BRI) countries
2023, Journal of Cleaner ProductionNetwork projection-based edge classification framework for signed networks
2020, Decision Support SystemsCitation Excerpt :Finally, we conclude the paper with future research directions in Section 7. Many data mining tasks can be performed on signed networks, such as node ranking, edge prediction, information diffusion, edge classification (sign/label prediction for edges), negative sign prediction for edges [18,25]. Among these tasks, in the present study, our focus is on the task of edge classification in signed networks which is a link-oriented task.
A robust trust inference algorithm in weighted signed social networks based on collaborative filtering and agreement as a similarity metric
2019, Journal of Network and Computer ApplicationsCitation Excerpt :Both theories are empowering many recent efforts. For example, Wang et al. (2015) provided a mathematical model for the status theory, and an algorithm for trust prediction in a graph where a link from a node u to another v means that u trusts v, while Yuan et al. (2017) developed a method using both theories for link sign prediction. It is also worth noting that experiments conducted by Tang et al. (2015) on datasets from Epinions and Slashdot show that more than 90% of the triads in these networks are consistent with both theories.
Implicit Negative Link Prediction With a Network Topology Perspective
2023, IEEE Transactions on Computational Social SystemsSigGAN: Adversarial Model for Learning Signed Relationships in Networks
2023, ACM Transactions on Knowledge Discovery from DataExponential bipartite synchronisation of stochastic signed networks via intermittent pinning discrete-time state observations control
2023, International Journal of Control
Weiwei Yuan is an associate professor in College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, China. She received her B.S. and M.S. degree from Harbin Engineering University, China in 2002 and 2005 respectively. She received her Ph.D. degree in Department of Computer Engineering, Kyung Hee University, South Korea in 2010. Her research interests include Pattern Recognition, Social Computing and Recommender Systems.
Chenliang Li is a master student in College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, China. He received his B.S. in College of Computer Science and Technology from China University of Mining and Technology (CUMT), Xuzhou, China in 2017. His research interests are data mining, machine learning, and social network analysis.
Guangjie Han is currently a Professor with the Department of Information and Communication System, Hohai University, Changzhou, China. He received the Ph.D. degree from Northeastern University, Shenyang, China, in 2004. From 2004 to 2006, he was a Product Manager for the ZTE Company. In February 2008, he finished his work as a Postdoctoral Researcher with the Department of Computer Science, Chonnam National University, Gwangju, Korea. From October 2010 to 2011, he was a Visiting Research Scholar with Osaka University, Suita, Japan. He is the author of over 220 papers published in related international conference proceedings and journals, and is the holder of 100 patents. His current research interests include sensor networks, computer communications, mobile cloud computing, and multimedia communication and security.
Donghai Guan is currently an associate professor with the College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing, China. He received the Ph.D. degree from Kyung Hee University (KHU), Suwon, Korea, in 2009. From 2009 to 2011, he was the research professor in Department of Computer Engineering, Kyung Hee University. From 2012 to 2014, he was the assistant professor in KHU. He has published more than 70 research papers in related international conferences and journals. His current research interests include machine learning, artificial intelligence recommender system, social network analysis, and ubiquitous computing.
Li Zhou is a master student in College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, China. She received her B.S. in College of Applied Science and Technology from Hainan University, China in 2015. Her research interests are Machine Learning, Data Mining and Social Network Analysis.
Kangya He is a master student in College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, China. He received his B.S. in College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, China in 2016. His research interests are Machine Learning, Data Mining and Social Network Analysis.