Influence learning for cascade diffusion models: focus on partial orders of infections

Probabilistic cascade models consider information diffusion as an iterative process in which information transits between users of a network. The problem of diffusion modeling then comes down to learning transmission probability distributions, depending on hidden influence relationships between users, in order to discover the main diffusion channels of the network. Various learning models have been proposed in the literature, but we argue that the diffusion mechanisms defined in most of these models are not well-adapted to deal with noisy diffusion events observed from real social networks, where transmissions of content occur between humans. Classical models usually have some difficulties for extracting the main regularities in such real-world settings. In this paper, we propose a relaxed learning process of the well-known independent cascade model that, rather than attempting to explain exact timestamps of users’ infections, focus on infection probabilities knowing sets of previously infected users. Furthermore, we propose a regularized learning scheme that allows the model to extract more generalizable transmission probabilities from training social data. Experiments show the effectiveness of our proposals, by considering the learned models for real-world prediction tasks.

[1]  Haiyan Wang,et al.  On the Existence of Positive Solutions of Fourth-Order Ordinary Differential Equations , 1995 .

[2]  Jure Leskovec,et al.  Inferring networks of diffusion and influence , 2010, KDD.

[3]  Ramanathan V. Guha,et al.  Information diffusion through blogspace , 2004, WWW '04.

[4]  Aram Galstyan,et al.  Information-theoretic measures of influence based on content dynamics , 2012, WSDM.

[5]  Masahiro Kimura,et al.  Generative Models of Information Diffusion with Asynchronous Timedelay , 2010, ACML.

[6]  Le Song,et al.  Scalable Influence Estimation in Continuous-Time Diffusion Networks , 2013, NIPS.

[7]  Éva Tardos,et al.  Maximizing the Spread of Influence through a Social Network , 2015, Theory Comput..

[8]  Jure Leskovec,et al.  Inferring Networks of Diffusion and Influence , 2012, ACM Trans. Knowl. Discov. Data.

[9]  Akshay Java,et al.  The ICWSM 2009 Spinn3r Dataset , 2009 .

[10]  Bernhard Schölkopf,et al.  Uncovering the Temporal Dynamics of Diffusion Networks , 2011, ICML.

[11]  Jure Leskovec,et al.  On the Convexity of Latent Social Network Inference , 2010, NIPS.

[12]  Hakim Hacid,et al.  A predictive model for the temporal dynamics of information diffusion in online social networks , 2012, WWW.

[13]  Masahiro Kimura,et al.  Learning Continuous-Time Information Diffusion Model for Social Behavioral Data Analysis , 2009, ACML.

[14]  Jure Leskovec,et al.  Meme-tracking and the dynamics of the news cycle , 2009, KDD.

[15]  Ludovic Denoyer,et al.  Learning social network embeddings for predicting information diffusion , 2014, WSDM.

[16]  Aristides Gionis,et al.  Sparsification of influence networks , 2011, KDD.

[17]  Jacob Goldenberg,et al.  Talk of the Network: A Complex Systems Look at the Underlying Process of Word-of-Mouth , 2001 .

[18]  Jon Kleinberg,et al.  Differences in the mechanics of information diffusion across topics: idioms, political hashtags, and complex contagion on twitter , 2011, WWW.

[19]  Bernardo A. Huberman,et al.  Predicting the popularity of online content , 2008, Commun. ACM.

[20]  Stefano Ermon,et al.  Feature-Enhanced Probabilistic Models for Diffusion Network Inference , 2012, ECML/PKDD.

[21]  Ludovic Denoyer,et al.  Predicting Information Diffusion in Social Networks Using Content and User's Profiles , 2013, ECIR.

[22]  Masahiro Kimura,et al.  Prediction of Information Diffusion Probabilities for Independent Cascade Model , 2008, KES.

[23]  Masahiro Kimura,et al.  Learning Diffusion Probability Based on Node Attributes in Social Networks , 2011, ISMIS.

[24]  Mark S. Granovetter Threshold Models of Collective Behavior , 1978, American Journal of Sociology.

[25]  Michael R. Lyu,et al.  Mining social networks using heat diffusion processes for marketing candidates selection , 2008, CIKM '08.

[26]  Yiming Yang,et al.  Introducing the Enron Corpus , 2004, CEAS.

[27]  Jure Leskovec,et al.  Modeling Information Diffusion in Implicit Networks , 2010, 2010 IEEE International Conference on Data Mining.