Author’s Accepted Manuscript

Behavior Analysis in Social Networks: Challenges,

Meng Wang, Ee-Peng Lim, Lei Li, Mehmet Orgun

www.elsevier.com/locate/neucom

PII: S0925-2312(16)30537-9
DOI: http://dx.doi.org/10.1016/j.neucom.2016.06.008
Reference: NEUCOM17146
To appear in: Neurocomputing
Accepted date: 3
Cite this article as: Meng Wang, Ee-Peng Lim, Lei Li and Mehmet Orgun,
Behavior Analysis in Social Networks: Challenges, Technologies, and Trends,
Neurocomputing, http://dx.doi.org/10.1016/j.neucom.2016.06.008
This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain.
Behavior Analysis in Social Networks: Challenges, Technologies, and
Trends
Meng Wang, Ee-Peng Lim, Lei Li, Mehmet Orgun
a
Hefei University of Technology
b
Singapore Management University
c
Hefei University of Technology
d
Macquarie University

The research on social networks has advanced significantly, which can be attributed
to the prevalence of the online social websites and instant messaging systems as well
as the popularity of mobile apps that support easy access to online social networks.
These social networks are usually characterized by the complex network structures
and rich contextual information. They now become the key platforms for, among
others, content dissemination, professional networking, recommendation, alerting,
and political campaigns. As online social network users perform activities on the
social networks, they leave data traces of human behavior which allow the latter to
be studied at scale. There are however a wide range of challenges in analyzing
human behavior in social networks. Behavior analysis in online social networks spans
a number of disciplines, across numerous fields in and beyond computer science. For
example, one would have to involve social network analysis, an area in social science,
to analyze social relationships, how they evolve and mature over time. The results of
behavior analysis have important implications on community discovery, anomaly
detection, and trend prediction, and they can enhance applications in multiple
domains such as information retrieval, recommendation systems, and trust and
security. Research in behavior analysis is a fertile ground also for businesses and IT
industry, as they develop innovative ideas fostering the design of the new generation
of social network platforms and their services.

This special issue is organized with the purpose of introducing novel research work
on behavior analysis in social networks. Submissions have come from an open call for
paper. With the assistance of professional referees, 26 papers are selected after at
least two rounds of rigorous reviews. These papers cover widely topics related to
behavior analysis in social network, including social network structure analysis, social
topic analysis and behavior understanding, social user analysis, social
recommendation, and social media analysis. We divide the whole special issue into
the above five parts according to the themes of the papers.
The first part contains 5 papers that are related to social network structure analysis.
In the first paper “Highly Efficient Epidemic Spreading Model Based LPA Threshold
Community Detection Method”, Deng et al. introduce an epidemic-spreading-based
label propagation process for community detection. The second paper,
“Multi-objective Community Detection Method by Integrating Users' behavior
attributes”, proposes a method that simultaneously accomplishes structure
clustering of social network and attribute categorization for user behaviors. The third
paper, “Uncovering Fuzzy Communities in Networks with Structural Similarity”,
proposes a fuzzy community detection method, which detects fuzzy community
structures without any prior knowledge. In the fourth paper, “Identification of
Influential Nodes in Social Networks with Community Structure based on Label
Propagation”, Zhao et al. propose a method that identifies a set of influential nodes
by exploring the community structure of social network. The fifth paper, “Discovering
top-k Non-Redundant Clusterings in Attributed Graphs”, introduces a novel algorithm
to discover the top-k non-redundant clustering solutions in attributed graphs, i.e., a
ranking of clusterings that share the least amount of information, in the information
theoretic perspective.

The second part contains 3 papers on social topic analysis and behavior
understanding. In the first paper “Predicting the Popularity of Viral Topics Based on
Time Series Forecasting”, Hu et al. demonstrate the high correlation of the
short-term popularity of viral topics, and present a method to predict the short-term
popularity of a given viral topic by using only data of historical popularity of the topic.
The second paper, “Estimating Multilateral Trade Behaviors on the World Trade Web
with limited information”, presents a novel approach to reconstructing the
topological structure of directed weighted network for estimating multilateral trade
behaviors. The third paper, “Protecting Private Geosocial Networks against Practical
Hybrid Attacks with Heterogeneous Information”, addresses the problem of privacy
preserving geosocial network data publishing and focuses on the anonymization to
prevent identity disclosure.

In the third part of the special issue, we have 6 papers on social user analysis. The
papers “Matching Entities across Online Social Networks” and “Identifying Users
Across Social Networks Based on Dynamic Core Interests” both discuss user
identification over multiple online social networks, one introduces a supervised
learning method with features extracted from users’ profiles and the other
introduces a dynamic core interests mapping method that jointly considers users’
social network structures and users’ generated content. The third paper, “FriendBurst:
Ranking People Who Get Friends Fast in a Short Time”, investigates a problem of
identifying the factors that are related to the burst of social users’ friends. It defines
the friend increasing speed ranking problem in a semi-supervised framework, and
then proposes a partially labeled ranking factor graph model to infer the ranking list
of friends increasing speed of users. In the fourth paper, “TOSI: A Trust-Oriented
Social Influence Evaluation Method in Contextual Social Networks”, Liu et al. propose
a trust-oriented social influence evaluation method built on three social contexts, i.e.,
social trust, social relationship and preference similarity, between two participants to
evaluate the social influence of each participant. In the fifth paper, “Identifying Social
Influence in Complex Networks: A Novel Conductance Eigenvector Centrality Model”,
Li et al. propose a conductance eigenvector centrality model to measure peer
influence in the complex social network. The sixth paper, “PT-LDA: A Latent Variable
Model to Predict Personality Traits of Social Network Users”, proposes a probabilistic
topic model to predict the personality traits within the framework of Five Factor
Model.

The fourth part of the special issue contains 5 papers on social recommendation. The
first paper, “An Intelligent Movie Recommendation System through Group-level
Sentiment Analysis in Microblogs”, introduces a movie recommendation approach
that mines user preferences information embedded in microblogs for evaluating the
similarity between online movies and TV episodes. The second paper, “Effective
Successive POI Recommendation Inferred with Individual Behavior and Group”,
introduces a point-of-interest recommendation method that combines the factors of
successive behaviors, group preference, and spatial restriction. In the third paper,
“Collaborative Filtering with Weighted Opinion Aspects”, Yang et al. propose a
collaborative filtering method that incorporates both user opinions and preferences
on different aspects. In the fourth paper, “SVM-TIA: A Shilling Attack Detection
Method Based on SVM and Target Item Analysis in Recommender Systems”, Zhou et
al. propose a method to detect shilling attacks based on SVM and target item analysis.
The final paper of this part, “Cluster-level Trust Prediction Based on Multi-modal
Social Network”, introduces a trust prediction approach that incorporates user-item
co-clustering method and the aggregation of multi-model similarity of users.

The final and also the largest part contains 7 papers on social media analysis. The
first two papers, “Sentiment Analysis for Chinese Microblog based on Deep Neural
Networks with Convolutional Extension Features” and “Topic-Related Chinese
Message Sentiment Analysis”, discuss sentiment analysis, one introduces a novel
deep learning approach and the other focuses on incorporating of topic information.
The second two papers, “Multi-Label Maximum Entropy Model for Social Emotion
Classification over Short Text” and “Detecting Influenza States based on Hybrid
Model with Personal Emotional Factors from Social Networks”, are related to social
emotion. In the fifth paper, “On the Learning of Image Social Relevance from
Heterogeneous Social Network”, Zhu et al. propose to decompose the heterogeneous
social network into several homogeneous networks, on which the global relevance
between social entities can be learned efficiently using a random walk style approach.
The sixth paper “Pornographic Image Detection Utilizing Deep Convolutional Neural
Networks” introduces a pornographic image detection using a convolutional neural
network. The seventh paper, “Dimensionality Reduction of Data Sequences for
Human Activity Recognition”, introduces a manifold elastic net that encodes the local
geometry for finding an aligned coordinate system in data representation.

To conclude, the papers in this special issue cover different techniques related to
social behavior analysis. We believe this special issue will benefit researchers and
practitioners working in this area.

Acknowledgement

We thank the reviewers for their great efforts. Their professional evaluations and
constructive comments are vital for securing the high quality of the special issue.
Finally, we express our thanks to all the authors who have contributed to the special
issue.