Identifying effective influencers based on trust for electronic word-of-mouth marketing: A domain-aware approach
Introduction
In the Internet era, advanced information technologies (such as personal computers, smart phones, wearable devices, and wireless broadband communications) and Internet-based social networking applications (such as Twitter, QQ, Facebook, and Skype) have innovated and empowered online social networks (OSNs). Today, the growing popularity of OSNs has changed the way that consumers and corporations interact with others. Consumers are accustomed to seeking useful information from people with a high online status in an OSN [25]; these people are normally referred to as “big Vs”, or more formally, influencers. Meanwhile, corporations are also making efforts to exploit the effects of influencers for sales and advertising [3], [34]. Thus, the ability to identify influencers in OSNs has become valuable to corporations in electronic word-of-mouth (eWOM) marketing, through which marketing information can be propagated faster and promoted better via recommendations by influencers in OSNs to their followers and peers [1], [12], [24]. In this way, e-commerce marketing campaigns that efficiently use social network resources not only reach more customers in OSNs but also increase their response rate.
As an important social concept, trust plays a critical role in users’ decisions, especially when the participants are anonymous and do not engage in direct face-to-face interactions [3], [29]. Users in an OSN expand their trust relationships with peers who have the same interests and preferences or with whom they have good interactions, and they thereby form a user trust network (UTN) [20]. According to Nielsen’s 2012 “Global Survey of Trust in Advertising”, 92% of consumers worldwide said that they trusted word-of-mouth recommendations from their trusted influential peers, vastly exceeding any other form of marketing, such as advertising or branded communications [2]. In this context, trusted influencers could potentially lead consumers to accept recommendations, make purchase decisions, and select transaction partners in e-commerce [44]. Therefore, identifying trusted influencers has become a popular research topic [20], [24], [45].
To date, numerous approaches to identifying influencers based on a UTN have been proposed [20], [24], [43], [45]. However, most studies have considered the effect of trust in an OSN to be universally applicable to all domains without domain-dependent specificity. Meanwhile, previous studies normally analyzed UTNs using static snapshots but neglected the nature of a UTN’s dynamics [20], [24]. These two issues may result in lower accuracy and efficiency for discovered influencers. For example, a comment regarding the taste of Fajita-Style Quesadillas written by an influencer who is well recognized in the Computer Hardware domain but not in the Home & Garden domain may not be convincing to his/her fans. In addition, this influencer may no longer be a highly regarded expert in the next time period in the Computer Hardware domain if he/she does not follow a new computer product, or his/her influential power may fade if rival influencers become more competent. Furthermore, corporations are not only interested in whether a user is currently an influencer but also in whether the user will maintain his/her influential power into the future. Therefore, it is important to re-address this research issue by proposing a novel approach reinforced by new dimensions.
This study focuses on identifying effective influencers by taking into account the dimensions of trust, domain, and time. In this paper, we develop a research framework based on social identity theory [30], [32] to identify effective influencers by combining a topological structure of domain-aware UTNs and the review time of users. First, we extend the traditional hypergraph to a time-varying hypergraph by adding in a time dimension to model a social network with multi-type relationship features and the dynamics of evolution. Second, we propose an algorithm to build a domain-aware UTN based on the hyperedge of the time-varying hypergraph and user trust relationships. Finally, we conceive a novel product review domain-aware (PRDA) approach to identify influencers by considering domain dependency and dynamics of trust, which overcomes the limits discussed above. According to their influential power status in their life cycle, we can define influencers as emerging influencers, holding influencers, and vanishing influencers. Based on the evolution of domain-aware UTNs, the PRDA approach can identify these three categories of influencers by using the social network analysis method. The experimental results reveal that many reviewers who could be defined as influencers using traditional approaches [20], [24], [43] never wrote any reviews in a specific product domain and hence have little influential power in that domain. This approach could be employed to identify effective influencers in a UTN in terms of domain and time period.
The contributions of our work can be summarized as follows: (1) based on social identity theory, we develop a research framework to identify effective influencers by combining review information and the trust relationships of users in OSNs; (2) a time-varying hypergraph is proposed to model social networks with multi-type relationship features and dynamic evolution; and (3) the PRDA approach is proposed to identify effective influencers based on the evolution of domain-aware UTNs.
The remainder of this paper is organized as follows. Section 2 introduces the theoretical background and related research. Then, the methodology and problem formalization is proposed in Section 3. Section 4 presents the empirical work and reports the evaluation results. The last section concludes the paper by summarizing the most important features of the proposed approach and suggesting future research directions.
Section snippets
Theoretical background and related research
There is a rich stream of literature examining social influence based on trust within OSNs for eWOM marketing purposes, particularly in e-commerce research. The existing studies were principally conducted based on three aspects of social networks: their theories, methods, and techniques. More specifically, these studies refer to social identity theory, online social trust, social network modeling, and influencer identification.
Methodology and problem formalization
In this section, we identify effective influencers by proposing a research framework that combines the structural properties of a UTN and the structured information of user reviews. The research framework is introduced in subsection 3.1; subsection 3.2 extends the traditional hypergraph to a time-varying hypergraph for modeling social networks; an algorithm for building a domain-aware UTN is proposed in subsection 3.3; and the novel product review domain-aware (PRDA) approach is presented in
Experiment design
To validate the proposed PRDA approach, we compared it with the commonly used social network-based influence-evaluating (SNIE) approach as well as the “popular author” approach. The existing SNIE approach that is applied to discover influencers does not consider the factors of specific domains and the dynamic evolution of the UTN [10], [20], [24], [43]. However, the “popular author” approach is a common online ranking method used to evaluate the influential power of influencers [23]. The reason
Conclusions
Identifying effective influencers in OSNs can be critical to save costs and create more business opportunities in eWOM marketing. Empowered by social identity theory, this study designs a comprehensive framework in which three dimensions – the trust relationship, review domains, and time – are used to identify influencers in an OSN. This research adopted the time-varying hypergraph to model social networks with the properties of multi-relationships and time dependency. Based on a domain-aware
Acknowledgements
The authors would like to thank the editor-in-chief, the associate editor, and four anonymous reviewers for their valuable comments and suggestions. This study was partially supported by a grant from the National Natural Science Foundation of China (Nos. 71331002, 71471054 and 61472057), and the Humanities and Social Sciences Fund Projects of the Ministry of Education (No. 13YJA630037).
References (45)
- et al.
A group trust metric for identifying people of trust in online social networks
Expert Syst. Appl.
(2012) - et al.
A theoretical model of intentional social action in online social networks
Decis. Support Syst.
(2010) - et al.
The impact of electronic word-of-mouth communication: a literature analysis and integrative model
Decis. Support Syst.
(2012) - et al.
Recommendation of similar users, resources and social networks in a Social Internetworking Scenario
Inf. Sci.
(2011) - et al.
Generating trusted graphs for trust evaluation in online social networks
Fut. Gener. Comp. Syst.
(2014) - et al.
Assessing the ripple effects of online opinion leaders with trust and distrust metrics
Expert Syst. Appl.
(2013) - et al.
Identification of influencers-measuring influence in customer networks
Decis. Support Syst.
(2008) - et al.
Discovering influencers for marketing in the blogosphere
Inf. Sci.
(2011) - et al.
Identifying influential reviewers for word-of-mouth marketing
Electron. Commer. Res. Appl.
(2010) - et al.
A graph-based action network framework to identify prestigious members through member’s prestige evolution
Decis. Support Syst.
(2012)
Leveraging social grouping for trust building in foreign electronic commerce firms: an exploratory study
Int. J. Inf. Manage.
Brand communication through digital influencers: leveraging blogger engagement
Int. J. Inf. Manage.
A social identity model of leadership effectiveness in organizations
Res. Organiz. Behav.
Improved trust-aware recommender system using small-worldness of trust networks
Knowl.-Based Syst.
User community discovery from multi-relational networks
Decis. Support Syst.
Discovering the influential users oriented to viral marketing based on online social network
Phys. A
Probabilistic subgraph matching on huge social networks
Crawling social internetworking systems
Models and Methods in Social Network Analysis
The SocialTrust framework for trusted social information management: architecture and algorithms
Inf. Sci.
Cited by (133)
LSS: A locality-based structure system to evaluate the spreader's importance in social complex networks
2023, Expert Systems with ApplicationsA social mechanism for healthcare consulting recommendation
2023, Information SystemsMicro, macro and mega-influencers on instagram: The power of persuasion via the parasocial relationship
2023, Journal of Business ResearchHow digital influencer content and characteristics influence Generation Y persuasiveness and purchase intention
2024, Tourism and Management StudiesScholar Influence Maximization via Opinion Leader and Graph Embedding Regression in Social Networks
2024, Communications in Computer and Information Science