Intelligent and Adaptive Systems Research Group Behavior Analysis

Intelligent and Adaptive Systems Research Group Behavior Analysis through Reputation Propagation in a Multi-Context Environment Ebrahim Bagheri and Ali A. Ghorbani E. Bagheri@unb. ca and Ghorbani@unb. ca Faculty of Computer Science - University of New Brunswick 12/18/2021

Intelligent and Adaptive Systems Research Group Reputation is a : Ødistributed, Øsocially ascribed, and Øcollective belief of the society towards a single identity. It can be ascribed to a Øsingle person, Øgroup, Ørole or even Øa non-human character. 12/18/2021 2

Intelligent and Adaptive Systems Research Group Reputation Ascription Reputation owners are affected by the status of their reputation. But have no control over the accredited reputation. Affecting reputation ascription through self alignment with social norms. 12/18/2021 3

Intelligent and Adaptive Systems Research Group Reputation Inheritance Assignment of reputation to a certain group of people, roles or objects A rough estimation of a typical group member’s reputation. This helps overcome reputation calculation for the unknown people to the society. 12/18/2021 4

Intelligent and Adaptive Systems Research Group Reputation Inheritance (cntd. ) Inheritance of reputation in a group setting (Dis)improvement of the status of the members based the group’s reputation Members affect the group’s reputation based on their social image. 12/18/2021 5

Intelligent and Adaptive Systems Research Group Reputation in Virtual Environments Interactions in a virtual environment based on reputation values. Interaction among users based on reputation and trust reputation is taken for granted since: Ø it has been communally calculated and ascribed. Reputation based systems should observe: ØIdentity Continuity and ØReputation Ubiquity. 12/18/2021 6

Intelligent and Adaptive Systems Research Group Identity Continuity & Reputation Ubiquity “Identity continuity” Øavoiding change of the identity or Øthe use of pseudonyms because: ØIt results in history clearance and Øfalse reputation declaration. “Reputation ubiquity” allows Øuniversal reputation values Øusers cannot hide their real identity 12/18/2021 7

Intelligent and Adaptive Systems Research Group Reputation Nuclei Knowledge Experience Relationship Role Fulfillment R Credential Faith Competence Honesty 12/18/2021 Favorability 8

Intelligent and Adaptive Systems Research Group Reputation Formalization Creating social norms based on these factors. Creation of context dependent reputation calculation models 12/18/2021 9

Intelligent and Adaptive Systems Research Group Reputation in a Single Context Reputation has been mostly formalized in a specific context. Ø Because of the underlying social values of each society For example in academia, for a prof. reputation is: Ø his/her publications, Ø teaching experience and Ø credentials; But in the business community: Ø industrial experience, Ø trading knowledge, and Ø Social relationships are more important than other factors. 12/18/2021 10

Intelligent and Adaptive Systems Research Group Mutual Effect of Different Contexts on Reputation is affected by the other contexts People with high reputation values in a specific context, will be looked upon as successful else where. The opposite situation is also true. Mutual effect of different contexts cannot be overlooked. 12/18/2021 11

Intelligent and Adaptive Systems Research Group Reputation in Multiple Context Knowledge Overall Reputation 1 2 3 4 Relationship Experience Credential Competence Honesty 12/18/2021 Role Fulfillment R Faith Favorability 12

Intelligent and Adaptive Systems Research Group Reputation Inference Understanding the position of a person in a new context People carry their reputation with them into new contexts This does not determine exactly the ultimate reputation of that person But it is a good initial estimation The person can later improve or damage his reputation 12/18/2021 13

Intelligent and Adaptive Systems Research Group Inter-context Reputation Migration We propose two steps for reputation migration: ØForward update mechanism ØBackward reputation adjustment method 12/18/2021 14

Intelligent and Adaptive Systems Research Group Forward Update Mechanism m 1 a h ec 2 e at M pd 3 rd U a rw 4 s ni Fo C 1 Reputation Change 12/18/2021 15

Intelligent and Adaptive Systems Research Group ec ha ni sm Backward Adjustment Mechanism tm en t. M 1 Ad ju s 2 4 Ba c kw ar d 3 NIC=8 NIC=4 NICi is calculated and shows the number of interactions that a person has in Context i. 12/18/2021 16

Intelligent and Adaptive Systems Research Group Scenarios Four Scenarios: ØConsistent Homogeneous Reputation ØHigh-Consistent Heterogeneous Reputation ØLow-Consistent Heterogeneous Reputation ØVarying Heterogeneous Reputation 12/18/2021 17

Intelligent and Adaptive Systems Research Group Observations 12/18/2021 18

Intelligent and Adaptive Systems Research Group Observations 12/18/2021 19

Intelligent and Adaptive Systems Research Group Conclusions Reputation is a context specific socially ascribed value. A specific reputation formalization cannot be applied to multiple contexts This prevents reputation inference in multiple contexts. Here we propose a reputation propagation scheme. The proposed model provides: Ø A rough anticipation of users’ behavior in new contexts Ø Analysis of reputation change in multiple contexts 12/18/2021 20

Intelligent and Adaptive Systems Research Group Thank you for your patience. 12/18/2021 21