Modelling Trust-based Evolutionary Dynamics in Signed Social Networks

Periodic Reporting for period 1 - TEAMS (Modelling Trust-based Evolutionary Dynamics in Signed Social Networks)

Reporting period: 2020-02-01 to 2022-01-31

The problem being addressed: This project focuses on detecting the trusted communities and capturing their evolutionary patterns over Signed Social Networks (SSNs) by using Formal Concept Analysis methodology.

Importance: Signed Social Network, a special form of online social network, can effectively portray human social network and has a wide range of applications in personalized recommendation, attitude prediction, user characteristic analysis and clustering, spam site identification, etc. Based on the evidences of practical applications in SSNs, the research results achieved in the project can serve as theoretic tools and potential technical solutions for supporting the development of trusted communities identification in SSNs to address critical issues related to European and global industries such as social marketing, journalism, and political science, and hence bring benefits for the society.

Overall objectives: This project aims to detect the trusted communities and capture their evolutionary patterns over SSNs. The specific objectives for the project “TEAMS” are listed as follows: (1) establish a unified representation model for a given signed social network; (2) establish the new model and develop innovative algorithms for detection of the trusted communities; (3) establish the evolutionary model and develop the innovative algorithms for the trusted communities; (4) systems evaluation, prototype system and demonstration for the trusted communities detection in SSNs.

WP1: Representation of Signed Social Networks
It regarded the vertex V in the signed social network as the set of objects and attributes with respect to the formal context in FCA, which is formalized as FC(G) = (V, V, I), I represents the positive and
negative relationship between nodes. And these relationships will be stored in an extended adjacency matrix M.

WP2-Investigation on Detection of Trusted Communities
In this WP, two efficient concept lattice generation algorithms Add-FCA and Dec-FCA, are developed for processing the incremental data. With these efficient algorithms, communities detection in social networks are investigated. These research results have been published in IEEE TNSE.
Y. Yang, F. Hao*, et.al: “Dynamic Maximal Cliques Detection and Evolution Management in Social Internet of Things: A Formal Concept Analysis Approach", IEEE Transactions on Network Science and Engineering, 2021.

Specifically, this project proved the equivalence between maximal clique and equiconcept. Further, a concept interestingness learning framework for identifying key topological structures from social networks is developed. These research results have been published in IEEE TNSE and GPC.

J. Gao, F. Hao*, et.al: “Learning Concept Interestingness for Identifying Key Structures from Social Networks", IEEE Transactions on Network Science and Engineering, Vol.8 No.4 pp.3220-3232 2021.
J. Gao, F. Hao*, et.al: “Concept Stability Based Isolated Maximal Cliques Detection in Dynamic Social Networks", The 15th International Conference on Green, Pervasive and Cloud Computing, pp.131–144, 2020.

In addition, this project utilized the three-way concept analysis methodology and stability of three-way concepts to address the structural issues in social networks.

F. Hao*, et.al: “Incremental Construction of Three-way Concept Lattice for Knowledge Discovery in Social Networks", Information Sciences, Vol.578 pp.257-280 2021.
F. Hao*, et.al: “Stability of Threeway Concepts and Its Application to Natural Language Generation", Pattern Recognition Letters, Vol.149 pp.51-58 2021.

Moreover, social media polarization problem is becoming increasingly serious, in order to model the multipolarization structure existing in the real signed networks, this project also developed a new cohesive subgraph model, called maximal multipolarized cliques based on the structure clustering theory. This research result has been published in ACM SIGIR 2021.

J. Gao, F. Hao*, et.al: “Maximal Multipolarized Cliques Search in Signed Networks", The 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, pp.2227–2231, 2021.

Considering the fuzzy property of social relationships, this project formulated a novel model of the skyline (λ, k)-cliques over a fuzzy attributed social network
and developed a Formal Concept Analysis (FCA) based skyline (λ, k)-cliques identification algorithm.

F. Hao*, et.al: “Skyline (λ, k)-cliques Identification from Fuzzy Attributed Social Networks", IEEE Transactions on Computational Social Systems, 2021.

WP3-Investigation on Evolutionary Dynamics of Trusted Communities
This WP mainly explores the evolution of maximal cliques in social networks byy observing different evolutionary patterns of maximal cliques including (1)unchanged maximal cliques; (2)changed maximal cliques; (3)added maximal cliques; (4) vanished maximal cliques, in social networks. To address it, this project described the evolution of social networks by analyzing the changes of several different categories of equiconcepts, when the nodes change.

Y. Yang, F. Hao*, et.al: “Dynamic Maximal Cliques Detection and Evolution Management in Social Internet of Things: A Formal Concept Analysis Approach", IEEE Transactions on Network Science and Engineering, 2021.

WP4-Systems Evaluation, Prototype System and Demonstration
This WP mainly builds up a personalized Ads recommendation system with the proposed algorithms in WP3. Technically, this WP defined the triadic timed formal concept analysis as a new methodology to solve the problem of location and context-aware advertisement recommendation on Twitter. These research results have been published in Soft Computing and IEEE SMC 2021.

C.De Maio, M. Gallo, F. Hao, et.al: “Who and where: Context-aware Advertisement Recommendation on Twitter", Soft Computing, Vol.25 pp.379–387, 2021.
C. De Maio, M. Gallo, F. Hao, et.al: “Fine-Grained Contextaware Ad Targeting on Social Media Platforms", The IEEE International Conference on Systems, Man, and Cybernetics, pp.3059-3065 2020.

The progress beyond the state of the art:
(1) This project developed two high-efficiency dynamical concept lattices generation algorithms, Add-FCA and Dec-FCA. Furthermore, a social-incremental concept lattice generation algorithm is developed.
(2) An equivalence theorem between maximal clique and equiconcept is presented and proved.
(3) A concept-cognitive learning framework for identifying key structures from social networks is presented by investigating the correlation between concept stability and key structures in social networks.
(4) An incremental three-way concepts generation algorithm is developed and the social networks oriented knowledge discovery approach is presented.
(5) This project explored the evolution of maximal cliques in social networks by observing different evolutionary patterns of maximal cliques in social networks.

The research outcomes achieved in this project develop a novel mechanism for understanding the topological structure formed by positive/negative links over social media, which set a milestone for future European and global studies and raise our visibility in the European academic and industrial circles. In addition to its importance in the wider scientific community, this research will also be of economic significance to related European and global industries such as social marketing, journalism, and political science.

Concept-cognitive learning framework for identifying key structures from social networks

Context-aware advertising recommendation system with TFCA

Periodic Reporting for period 1 - TEAMS (Modelling Trust-based Evolutionary Dynamics in Signed Social Networks)

Share this page

Download