Modeling the rational behavior of buyers and sellers in an E-commerce system using agent-based simulation

Document Type : Original Article

Authors

1 Department of Industrial Engineering, Yazd University, Yazd, Iran

2 Department of Industrial Engineering, Faculty of Engineering, Semnan University, Semnan, Iran

Abstract

In today’s world, the popularity of E-commerce and online transactions has increased dramatically. Such transactions are done indirectly with the help of an intermediary or a third party. A review of the literature shows that reputation plays an important role in reducing buyer’s risk in such situations. An online reputation management system is a system that is formed from the positive or negative beliefs or opinions of people about a person, product or service and can guarantee the reliability of transactions in the E-commerce systems. In these systems, shipping services and transportation costs are also among the items that can influence the buyer’s decision, because like the traditional way, the buyer does not buy the product from the store or in person, and factors such as the distance between buyer and seller might impact the shipping cost or even buyer’s trust. Buyer-seller interaction is a traditional field of application of game theory. In this research, the combined approach of agent-based simulation and game theory is used to evaluate an E-commerce system, in which a centralized reputation management system is provided by a trusted third party. Then, a numerical evaluation is performed to evaluate the proposed model, in which the impact of the policies of a group of buyers and sellers and their behavior in the E-commerce system is modeled.

Keywords

Main Subjects


Abstract

In today’s world, the popularity of E-commerce and online transactions has increased dramatically. Such transactions are done indirectly with the help of an intermediary or a third party. A review of the literature shows that reputation plays an important role in reducing buyer’s risk in such situations. An online reputation management system is a system that is formed from the positive or negative beliefs or opinions of people about a person, product or service and can guarantee the reliability of transactions in the E-commerce systems. In these systems, shipping services and transportation costs are also among the items that can influence the buyer’s decision, because like the traditional way, the buyer does not buy the product from the store or in person, and factors such as the distance between buyer and seller might impact the shipping cost or even buyer’s trust. Buyer-seller interaction is a traditional field of ​​application of game theory. The use of game theory in this field is relatively limited due to its complex solutions. One way to deal with such complexities is to use agent-based simulation capabilities. In this research, the combined approach of agent-based simulation and game theory is used to evaluate an E-commerce system, in which a centralized reputation management system is provided by a trusted third party. Then, a numerical evaluation is performed to evaluate the proposed model, in which the impact of the policies of a group of buyers and sellers and their behavior in the E-commerce system is modeled. Considering the location parameter for individuals, we added shipping costs to the model. As expected, the number of transactions between the buyer and the non-citizen seller decreased. To increase the number of these transactions, we added a discount policy in which, in case the buyer buys the products more than a certain price, the shipping cost is free for him/her. Then, the optimal discount price for this policy was obtained and validated with the use of sensitivity analysis, in order to increase the number of transactions between the buyers and the non-citizen sellers.

Keywords: E-commerce, Reputation management, Agent-based modeling and simulation, Game theory, Shipping cost.

 

Introduction

Despite the increasing growth of e-commerce systems, evidence shows that buyers' trust in online shopping is lower than in-person shopping, and buyers are more difficult to trust online shopping. Studies have shown that reputation plays a very important role in reducing buyer risk in an e-commerce environment (Aringiri et al., 2017). Reputation is not only the result of the personal and direct experience of the buyer, but also includes any other type of communication, such as points or ratings and surveys, which include information about the seller (Lappas et al., 2016). An online reputation system consists of beliefs and opinions about a person or thing, which is a solution to guarantee transaction trust in an online business environment. Since in online business, shopping is done without going to the store, the buyer does not see the product closely and may have to pay for the delivery of the product and receive the product with a delay. In particular, regarding delivery, the buyer may face problems such as: delivery of goods at the time of absence, delivery delays, high delivery costs, and damaged goods. Interactions between buyers and sellers have been studied in the game theory approach for a long time. In this article, the game theory approach has been used to model the rational behavior of individuals and the reputation management system. It is worth noting that due to the complexity of the relevant solution methods, the use of this approach has been limited in practice (Aringiri et al., 2017). It should be noted that such complexity is not related to a single transaction, but depends on the fact that reputation is the result of two things: (1) the execution of a number of transactions between a number of buyers and sellers, which may not necessarily be the same, and (2) the sharing of transaction results with other buyers and sellers that this sharing demonstrates the impact of learning in repeatable games. A solution to deal with such complexities is to use agent-based simulation capabilities (Gilbert, 2008). In this approach, a set of rules, the behavior of agents and their interactions with the environment are specified, which determines the state and position of each agent (Aringiri et al., 2017). With these conditions, the behavior of the agents can be investigated and the necessary information can be extracted.

In online business, the time of sending the goods, the method and cost of sending, and in general all other issues that affect the transaction from the moment of confirmation of the purchase by the buyer till the receipt of the goods are very important. It is believed that the shipping cost has an effect on the buyer's decision to choose a seller and makes the buyer less willing to buy from online sellers and among them from non-local sellers. Because otherwise, it will require paying more shipping fees. Also, the behavior of sellers to their fellow-citizen buyers is different. Because in the case of fraud in the delivery of the goods by the seller, the buyer has an easier way to follow up. Therefore, we are looking for a policy evaluation that will reduce such effects on the internet competition market.

 

Methodology

The basic model of the online trading can be represented by a two-player game consisting of a buyer (B) and a seller (S). Considering only one transaction for a product from seller to buyer, the buyer should first choose between purchasing the product (P) or refusing the purchase (R). Then, if the buyer decides to purchase the goods (P), the seller should decide whether to deliver the goods to the customer (D) or keep both the money and the goods for himself (K). These conditions lead to the creation of a non-cooperative game with the set of players , the set of buyer's strategies , and the set of seller's strategies .

The basic agent-based simulation model consists of three agents: buyer, seller, and observer. Buyers and sellers are placed in the "small world" network environment. In this type of network, each agent has the possibility of communication with another agent. Since the buyer's needs may change over time, in each time period , the relationships of each buyer can change and be different from other times. In this network, only a number of sellers can provide the goods in a certain period of time because other sellers may be out of stock.

In each time interval , the buyer inquires the reputation value of one of the related sellers from the observer. After calculating the reputation value, the buyer decides to trust the seller and complete the transaction or wait until the next time interval. When a transaction is completed, the buyer provides feedback of this transaction to the observer to update the seller's reputation. The buyer's personal experience will also be updated.

 

Findings

The effect of the goods delivery system on the behavior of the entire online trading system is examined. The shipping cost is calculated on the amount paid by the buyer and affects her decision to choose the seller. The results show that considering the cost of transporting the goods, buyers have shown less desire to buy from non-local sellers. The findings show that the behavior of individuals with the introduction of the delivery system is still reasonable and close to reality. It can be seen that by adding the product discount policy, the number of transactions can be increased and the adverse effect of shipping costs can be reduced. The findings show that in e-commerce, it is possible to reduce the effect of shipping costs for the customer with discount policies so that the customers can take advantage of more options in their choice, and a more complete competition market is created for the sellers.

 

Conclusion and recommendations

In this research, a hybrid model including game theory and agent-based simulation was presented to evaluate the online trading system by considering reputation. Game theory has been used to create the rational behavior of buyers and sellers, and agent-based simulation has been used to model the entire online trading system and its infrastructure network. Such an approach solves the complexity of the game theory approach. A coherent numerical evaluation was presented in order to validate the model and to investigate the effect of a set of policies for buyers and sellers, and a discount policy for shipping goods on the online trading system.

For future researches, it is suggested to develop the presented model from two perspectives of modeling and implementation. As an example, from the perspective of modeling, we can consider a market where players can play both the role of seller and buyer at different times. From the operational point of view, the inherent flexibility of the proposed model approach to evaluate the effectiveness of a reputation management system with regard to trust management in cloud computing can be investigated.

 

References

Aringhieri, R., Duma, D. and Fragnelli, V. “Modeling the rational behavior of individuals on an e-commerce system”, Operations Research Perspectives, 5, pp.22-31 (2018).

Gilbert, N. “Agent-Based Models (Quantitative Applications in the Social Sciences)”, SAGE Publications (2008).

Lappas, T., Sabnis, G. and Valkanas, G. “The impact of fake reviews on online visibility: A vulnerability assessment of the hotel industry”, Information Systems Research, 27(4), pp.940-961 (2016).

Aringhieri, R., Duma, D. and Fragnelli, V. “Modeling the rational behavior of individuals on an e-commerce system”, Operations Research Perspectives, 5, pp.22-31 (2018).
Borshchev, A. and Filippov, A. “From system dynamics and discrete event to practical agent based modeling: reasons, techniques, tools”, In Proceedings of the 22nd international conference of the system dynamics society, 22, pp. 25-29 (2004).
Čavoški, S. and Marković, A. “Agent-based modelling and simulation in the analysis of customer behaviour on B2C e-commerce sites”, Journal of Simulation, 11(4), pp.335-345 (2017).
Cioppi, M., Curina, I., Forlani, F. and Pencarelli, T. “Online presence, visibility and reputation: a systematic literature review in management studies”, Journal of Research in Interactive Marketing, 13(4), pp. 547-577 (2019),.
Gilbert, N. “Agent-Based Models (Quantitative Applications in the Social Sciences)”, SAGE Publications (2008).
Hantula, D.A. and Bryant, K. “Delay discounting determines delivery fees in an   e-commerce simulation: A behavioral economic perspective”, Psychology & Marketing, 22(2), pp.153-161 (2005).
Jiang, G., Liu, S., Liu, W. and Xu, Y. “Agent-based modeling and simulation of the decision behaviors of e-retailers”, Industrial Management & Data Systems (2018).
Koyuncu, C. and Bhattacharya, G. “The impacts of quickness, price, payment risk, and delivery issues on on-line shopping”, The Journal of Socio-Economics, 33(2), pp.241-251 (2004).
Lappas, T., Sabnis, G. and Valkanas, G. “The impact of fake reviews on online visibility: A vulnerability assessment of the hotel industry”, Information Systems Research, 27(4), pp.940-961 (2016).
Morganti, E., Seidel, S., Blanquart, C., Dablanc, L. and Lenz, B. “The impact of e-commerce on final deliveries: alternative parcel delivery services in France and Germany”. Transportation Research Procedia, 4, pp.178-190 (2014).
Niu, W. “Research on E-Commerce Customer Feature Extraction Question Answering System Based on Artificial Intelligence Semantic Analysis”. Advances in Multimedia, 2022, pp. 1687-5680 (2022).
Riazati, M., Shajari, M. and Khorsandi, S. “An incentive mechanism to promote honesty among seller agents in electronic marketplaces”, Electronic Commerce Research, 19(1), pp.231-255 (2019).
Xie, H., Ma, R.T. and Lui, J.C. “Enhancing reputation via price discounts in e-commerce systems: A data-driven approach”, ACM Transactions on Knowledge Discovery from Data (TKDD), 12(3), pp.1-29 (2018).
Xu, M., Ferrand, B. and Roberts, M. “The last mile of e-commerce–unattended delivery from the consumers and eTailers' perspectives”, International Journal of Electronic Marketing and Retailing, 2(1), pp.20-38 (2008).
Wu, Y., Xue, C., and Li, S. “Optimization Analysis of Customs Supervision Strategy Based on Complex Network Evolutionary Game”, Proceedings of the 2022 International Conference on Bigdata Blockchain and Economy Management, pp.658-668 (2022).