Link Prediction in Aligned Heterogeneous Networks
Lecture notes in computer sciencePublished 1 January 2015
Fangbing Liu, Shu‐Tao Xia
Citations7
SJR quartileQ2
SJR score0.35
SNIP0.55
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
This paper attempts to solve the user-user link prediction problem for new users by utilizing data in a similar social network (source network) and proposes the Aligned Factor Graph (AFG) model, which works well when users leave little data in target network.
Abstract
Social networks develop rapidly and often contain heterogeneous information. When users join a new social network, recommendation affects their first impressions on this social network. Therefore link prediction for new users is significant. However, due to the lack of sufficient active data of new users in the new social network (target network), link prediction often encounters the cold start problem. In this paper, we attempt to solve the user-user link prediction problem for new users by utilizing data in a similar social network (source network). In order to bridge the two networks, three categories of local features related to single edge and one category of global features associated with multiple edges are selected. The Aligned Factor Graph (AFG) model is proposed for prediction, and Aligned Structure Algorithm is used to reduce the factor graph scale and keep the prediction performance at the same time. Experiments on two real social networks, i.e., Twitter and Foursquare show that AFG model works well when users leave little data in target network.
Keywords
Computer SciencePhysics and Astronomy
Signed networks in social media
1,287 Citations2010Jure Leskovec, Daniel P. Huttenlocher +1 more
This work provides one of the first large-scale evaluations of theories of signed networks using on-line datasets, as well as providing a perspective for reasoning about social media sites.
Supervised random walks
996 Citations2011Lars Bäckström, Jure Leskovec
An algorithm based on Supervised Random Walks is developed that naturally combines the information from the network structure with node and edge level attributes and outperforms state-of-the-art unsupervised approaches as well as approaches that are based on feature extraction.
Modeling relationship strength in online social networks
664 Citations2010Rongjing Xiang, Jennifer Neville +1 more
This work develops an unsupervised model to estimate relationship strength from interaction activity and user similarity and evaluates it on real-world data from Facebook and LinkedIn, showing that the estimated link weights result in higher autocorrelation and lead to improved classification accuracy.
Exploiting place features in link prediction on location-based social networks
528 Citations2011Salvatore Scellato, Anastasios Noulas +1 more
A supervised learning framework is described which exploits new prediction features based on the properties of the places visited by users which are able to discriminate potential future links among them and which exploits these prediction features to predict new links among friends-of-friends and place-friends.
Europhysics Letters (EPL)Link prediction in weighted networks: The role of weak ties
316 Citations2010Linyuan Lü, Tao Zhou
This letter uses local similarity indices to estimate the likelihood of the existence of links in weighted networks, including Common Neighbor, Adamic-Adar Index, Resource Allocation Index, and their weighted versions, and gives a semi-quantitative explanation based on the motif analysis.
Inferring social ties across heterogenous networks
300 Citations2012Jie Tang, Tiancheng Lou +1 more
The framework incorporates social theories into a factor graph model, which effectively improves the accuracy of inferring the type of social relationships in a target network by borrowing knowledge from a different source network.
Cross-domain collaboration recommendation
293 Citations2012Jie Tang, Sen Wu +2 more
The Cross-domain Topic Learning (CTL) model is proposed, which consolidates the existing cross-domain collaborations through topic layers instead of at author layers, which alleviates the sparseness issue and outperforms baselines significantly on multiple recommendation metrics.
Link Prediction and Recommendation across Heterogeneous Social Networks
281 Citations2012Yuxiao Dong, Jie Tang +5 more
This work gives a formal definition of link recommendation across heterogeneous networks, and proposes a ranking factor graph model (RFG) for predicting links in social networks, which effectively improves the predictive performance.
Translated Learning: Transfer Learning across Different Feature Spaces
266 Citations2008Wenyuan Dai, Yuqiang Chen +3 more
Through experiments on the text-aided image classification and cross-language classification tasks, it is demonstrated that the translated learning framework can greatly outperform many state-of-the-art baseline methods.
Mining topic-level influence in heterogeneous networks
249 Citations2010Lu Liu, Jie Tang +3 more
A generative graphical model is proposed which utilizes the heterogeneous link information and the textual content associated with each node in the network to mine topic-level direct influence and a topic- level influence propagation and aggregation algorithm is proposed to derive the indirect influence between nodes.
Link Prediction in Social Networks Using Computationally Efficient Topological Features
184 Citations2011Michael Fire, Lena Tenenboim +4 more
This paper proposes a set of simple, easy-to-compute structural features that can be analyzed to identify missing links and shows that a machine learning classifier trained using the proposed simple structural features can successfully identifyMissing links even when applied to a hard problem of classifying links between individuals who have at least one common friend.
Cold start link prediction
165 Citations2010Vincent Leroy, B. Barla Cambazoğlu +1 more
This paper introduces cold start link prediction as the problem of predicting the structure of a social network when the network itself is totally missing while some other information regarding the nodes is available.
Bethe free energy, Kikuchi approximations, and belief propagation algorithms
164 Citations2001Jonathan S. Yedidia, William T. Freeman +1 more
It is shown that BP canonly converge to a stationarypoint of an approximatefree energy, known astheBethefreeenergy in statisticalphysics, as of its release on April 4, 2001.
Cross-Domain Recommender Systems
146 Citations2011Paolo Cremonesi, Antonio Tripodi +1 more
This paper formalizes the cross-domain problem in order to provide a common framework for the classification and the evaluation of state-of-the-art algorithms and defines a new class of cross-domain algorithms based on neighborhood collaborative filtering, either item-based or user-based.
Information Systems FrontiersSocial balance in signed networks
105 Citations2014Xiaolong Zheng, Daniel Zeng +1 more
A brief survey of the existing balance measures and present detecting/partitioning algorithms, as well as important empirical investigations in both physical world and cyberspace are discussed.
Predicting Social Links for New Users across Aligned Heterogeneous Social Networks
104 Citations2013Jiawei Zhang, Xiangnan Kong +1 more
A link prediction method called SCAN-PS (Supervised Cross Aligned Networks link prediction with Personalized Sampling), to solve the social link prediction problem for new users and could solve the cold start problem when information of these new users is total absent in the target network.
Evolution of a location-based online social network
79 Citations2012Miltiadis Allamanis, Salvatore Scellato +1 more
It is demonstrated that geographic distance plays an important role in the creation of new social connections: node degree and spatial distance can be combined in a gravitational attachment process that reproduces real traces, and a new model of network growth is described that combines spatial and social factors.
Factor Graphs and Algorithms
76 Citations2008Brendan J. Frey, Frank R. Kschischang +1 more
A general algorithm for computing marginals of the global function by distributed message passing in the corresponding factor graph is described.
Social Network Analysis and MiningSupervised methods for multi-relational link prediction
72 Citations2012Darcy Davis, Ryan N. Lichtenwalter +1 more
A novel probabilistically weighted extension of the Adamic/Adar measure for heterogenous information networks is introduced, which is used to demonstrate the potential benefits of diverse evidence, particularly in cases where homogeneous relationships are very sparse.
Selective Transfer Learning for Cross Domain Recommendation
58 Citations2013Zhongqi Lu, Erheng Zhong +4 more
This paper proposes a novel criterion based on empirical prediction error and its variance to better capture the consistency across domains in CF settings and embeds this criterion into a boosting framework to perform selective knowledge transfer.