Multi-agent single-objective negotiation mechanism of personalized product supply chain based on personalized index

Abstract

In the multi-agent system of personalized product supply chain, negotiation can effectively solve conflict and realize cooperation. In this article, personalized index is defined and the calculation method of personalized index is presented. Then, the multi-agent negotiation model is constructed based on personalized index and the negotiation procedure is discussed. Finally, the validity of the model is demonstrated by an example. That will be provided theoretical and operational methods for personalized product supply chain management, is helpful to coordinate and control personalized product supply chain, and realize the efficient, flexible, and quick operation of personalized product supply chain.

Keywords

Artificial intelligence multi-agent negotiating mechanism negotiating model and tactics personalized index personalized product supply chain management

Introduction

With the continuous improvement of living standards, customer demands have changed gradually. The change shows that customers are not satisfied with standardized product and prefer to meet the personalized product.^1,2 The ability of meeting the personalized needs of customers has become an indispensable ability for manufacturing enterprises. Only by designing, producing, and serving different customers, the core competitiveness of manufacturing enterprise can be improved. In the demand-driven market, it is the key point for manufacturing enterprise to understand the customer’s individual demand accurately and adopt the appropriate supply chain model to respond quickly to the personalized requirements of customers.^3–5

The research of manufacturing enterprise personalized product focuses on personalized product requirements and personalized product customization.^6–9 Personalized product requirement research mainly includes the characteristics of personalized demand and concepts, demand forecasting, and product classification.^10–15 Personalized product customization research mainly includes the concept and implementation methods of personalized customization.^12–18 The personalized product supply chain refers to a customer-oriented supply chain mode. This mode takes customer’s personalized demand as the starting point, integrates customer, production integrators, suppliers, logistics service providers, and other service providers by utilizing artificial intelligence, big data, cloud computing, Internet, and Internet of things technology; this mode integrates supply chain resources to provide personalized product for customer by means of the analysis of demand forecasting and collaborative operation, and focuses on customer’s personalized experience and maximum satisfaction. These entities can be abstracted to the independent agent, and personalized product supply chain can be considered as a multi-agent system composed of agent, which is divided into separate enterprises or departments. In the operation of personalized product supply chain, negotiation can effectively solve the conflict between partners, and it is also an important way to improve the operational efficiency of personalized product supply chain. At present, research on personalized product supply chain and negotiation of personalized product supply chain is rare.

Research on agent and multi-agent systems has originated from distributed artificial intelligence, which is the inevitable result of the development of distributed artificial intelligence, modern computer technology, and communication technology.^19–21 Agent and multi-agent systems are frequently used to solve the problems of decision-making and collaboration in supply chain management. Shoham²² believes that if the state of an entity contains the mental states of belief, commitment, knowledge, and ability, then the entity is an agent. Wooldridge and Jennings suggest that the agent is a computer system to meet the needs of specific design, and it has superior flexibility and autonomy.^23,24 Franklin and Graesser²⁵ believe that the agent is a software entity that can represent the user, has certain independence and autonomy, and automatically realizes the original design of a set of targets in the calculation of complex dynamic system environment. A multi-agent system is composed of multiple agents that coordinate with each other to accomplish their goals. In view of the limited ability of a single agent to solve complex problems, it is possible to form a multi-agent system by coordinating multiple independent agent behaviors to compensate for the lack of a single agent function. In a multi-agent system, each agent coordinates each other’s goals and completes a specific task or achieves goals through cooperation.

Research on multi-agent systems is concentrated on multi-agent model, architecture, consistency and coordination, multi-agent planning, and conflict processing.²⁶ The negotiation process can be seen as a process in which agents coordinate their respective actions through communications to take joint action. Multi-agent communication technology is a fundamental foundation and service guarantee for negotiation. KQML is an officer communication language based on speech act theory, which has become the fact standard of agent communication language.²⁷ Implementing agent negotiation in addition to agreeing on communication languages, there is a need for a negotiation mechanism to regulate the collaboration process. Contract net protocol (CNP) is part of the most well-known and widely used consultative mechanisms.^28,29

Negotiation is the exchange of relevant structured information by both parties, forming planning and consensus of views.³⁰ Neumann and Morgenstern³¹ divide the negotiation into single-objective and multi-objective negotiation. Faratin et al.³² believe that negotiation can be divided into time-, resource-, and behavior-dependent negotiation from the perspective of negotiation strategy, and point out that the main factors affecting the negotiation process convergence include time, resources, and behavior and also that the deliberative bodies can produce a variety of negotiation tactics based on these factors. Jennings et al.³³ studied the division method of negotiation research category and considered that the scope of negotiation research mainly includes negotiation language category, decision-making category, and process category. Kraus³⁴ believes that negotiation is an important link in deciding whether an electronic transaction is successful. He proposes a general model of multi-agent negotiation framework and discusses multi-agent negotiation under incomplete information. The model mainly discusses the single-objective negotiation between the two sides.^34,35 Faratin et al.³² put forward the computational function of fuzzy values based on agent negotiation problem. The validity of the calculation function is verified by experiment and simulation. Jennings et al.³³ developed a method of calculating the price utility function by fuzzy mathematics, and the price negotiation model and tactics based on multi-agent system are discussed. Fatima studied the negotiation of price and utility based on web between the two negotiating parties. A multi-agent negotiation model and tactics for two commodity prices under incomplete information are proposed.^36,37 Park and Yang³⁸ propose a multi-objective negotiation model for e-commerce environment and discussed the ways to make the negotiation agent reach the satisfactory or optimal solution. Li and Sheng³⁹ discuss the multi-agent negotiation model under uncertainty information. Huang and Liao⁴⁰ propose a B2C e-commerce negotiation model including information collection, search, negotiation, and evaluation. Adhau proposes a distributed multi-agent system by means of auction based on negotiation method for solving multi-project schedule problem. Computational experiments showed the impact of resource transfer time for project delay and can minimize the project duration.⁴¹ Chen and Weiss⁴² presented an automated negotiation tactics that can adjust utility based on adaptive concession-making mechanism by acquiring an opponent model and provided an empirical analysis based on game theory that showed the robustness of the automated negotiating strategy. Hernández et al.⁴³ discuss the supply chain negotiation mechanism and the supply chain coordination planning based on multi-agent system. Khaled and Besoa⁴⁴ discuss the supply chain performance based on agent simulation. The application of agent will increase the efficiency of supply chain operation. Mansour considered a one-to-many negotiation approach of the buyer agent, addressed the bidding tactics in the negotiation process, and proposed novel dynamic negotiating tactics. The experimental results approved the effectiveness of dynamic negotiating tactics in different negotiating environments.⁴⁵ Manupati et al.⁴⁶ describe a negotiation approach–based mobile agent and presented the functions and the fundamental framework of the method, and an illustrative example was presented to validate the feasibility of the approach. Ilany and Gal⁴⁷ discussed how to select algorithm in bilateral negotiation and designed a meta-agent for predicting the performance of different negotiating tactics; the simulation results showed the effectiveness of the method.

The above works provide lots of methods for solving the negotiation problem and application of agent and multi-agent systems in supply chain negotiation. For example, Neumann and Morgenstern,³¹ Faratin et al.,³² and Jennings et al.³² studied the multi-agent negotiation categories and influencing factors. Kraus and colleagues^34,35 explore the multi-agent negotiation model of a single objective under incomplete information conditions. And in some previous works,^{32,33,36–39} the agent negotiation problem was explored using fuzzy mathematics method and also the calculation method of negotiation utility function and the offer tactics of negotiation parties were explored. Although these works reported the negotiation problems from a different aspect, the influence of different risk preferences on the negotiation process is not taken into account, and the negotiation in personalized product supply chain is also worth studying. However, in the actual negotiation process, different risk preferences will affect the negotiation number, price, and utility, and the purpose of this study is to explore the impact of different risk preferences on the negotiation process in personalized product supply chain. Based on the above research, we will analyze the personalized index and then build the bidding tactics model of both parties, which take into account the impact of different risk preferences and discuss how different risk preferences affect the negotiation number, price, and utility by experiment. We discuss the personalized index in section “Discussion of personalized index.” We present the bidding tactics model and analyze the negotiation procedure in section “Negotiation model and tactics.” In section “Negotiation illustrations,” a concrete example will be given for analyzing the influence of different risk preferences on the negotiation process, which can provide the method for solving agent negotiation in personalized product supply chain and demonstrate the validity and correctness of the models. Finally, we present the conclusion and future work.

Discussion of personalized index

Personalized index

In the multi-agent negotiation model of personalized product supply chain based on personalized index, the personalized demand of consumer has an important impact on the negotiation process. In order to reflect this influence, this article proposes a measure of personalized degree, that is, personalized index. Therefore, this article assumes that the personalized degree in the negotiation model can be measured by personalized index, where g is used to represent the personalized index. The personalized index g is a relative concept and is relative to the standardized product. Here we define the standardized product; for any product, it meets the basic functions of customer need. There are lots of products in the market, and have no difference among the same product from materials, quality, functional process, structure, and other aspects. And if g = 1, it means the standardized product. There are three kinds of values of g which are shown as follows

g = {\begin{matrix} (0, 1) \\ 1 \\ (1, + \infty) \end{matrix}

(1)

The first value is that, compared with the standardized product, when a product removes certain characteristics or a feature for the general user to become worse, such as function, materials, quality, and some product characteristics without additional features, the personalized index is in the interval (0,1). For example, some customers do not want to choose the best supplier to provide parts for their personalized products, but choose suppliers with low price.

The second value is when the product is a standardized product, that is, all the same products in the market are exactly the same, and the personalized index is 1.

The third value is that, compared with the standardized product, when product characteristic for the general user becomes better, such as more attractive appearance and more diverse function, and the best suppliers for providing parts for their personalized products, the interval of the personalized index is (1,+∞).

Impact of personalized index

The influence of the personalized index on the buyer’s offer can be represented by the following function.

The offer of the buyer is as follows

x = g p_{0}

(2)

where x is the offer of the buyer, g is the personalized index, and $p_{0}$ is the offer of the buyer for a standardized product, which is the offer of the buyer when the personalized index is 1. The personalized index of the buyer’s offer is a linear relationship. The function graph is shown in Figure 1(a).

Figure 1.

(a) Buyer’s quotation and (b) seller’s quotation.

The impact of the personalized index on the seller’s offer can be represented by the following function, and the function graph is shown in Figure 1(b).

The offer of the seller is as follows

y = p_{0} \log_{2} (g + 1)

(3)

where y is the offer of the seller, g is the personalized index, and $p_{0}$ is the offer of the seller for a standardized product, which is the offer of the seller when the personalized index is 1.

Hence, the offer of the seller consists of three kinds

y = {\begin{matrix} (0, p_{0}), & 0 < g < 1 \\ p_{0}, & g = 1 \\ (p_{0}, + \infty), & g > 1 \end{matrix}

(4)

When the product’s personalized index is in the range (0,1), the product characteristics will be reduced compared with the standardized product. The cost of the enterprise is reduced, so the offer of the seller can be lower than the standardized product, that is, less than $p_{0}$ .

When the value of a product’s personalized index is 1, the product is a standardized product and the offer of the seller is $p_{0}$ .

When the value of a product’s personalized index is greater than 1, it indicates that the characteristics of a product have improved compared with the standardized product, which may be an increase in function or improvement of the material. The offer of the seller is greater than $p_{0}$ .

It can be seen from Figure 1(b) that the offer of the seller is an increasing function of the personalized index g and is a convex function. When a product’s personalized index g is in the range from 0 to 1, the cost of the product is reduced compared to a standardized product; as the product characteristics decrease, the cost reduction will be greater and greater, and each unit characteristic decreases, which is determined by the product integrator’s self-learning. As the personalized needs are still to be met, product integrators for the production of personalized product become more and more sophisticated and the cost reduction ratio is also growing.

When a product’s personalized index g is more than 1. Compared with the standardized product, the cost of the product will increase for each additional feature. Moreover, with the increase in product characteristics, the increase in the cost per unit of product characteristics caused by the increase will be less important and smaller. This is because when the product is compared with a standardized product, one unit of property is added, the producer is required to make a drastic adjustment from the scale of production, and therefore the added cost is large. Relatively speaking, with the characteristics of the increase, production adjustment of producers is not just beginning so significant, so the increase in cost is smaller.

Negotiation model and tactics

Negotiation model

The multi-agent single-objective negotiation of the personalized product supply chain based on personalized index is the negotiation between the product integrator and the consumer. The negotiation of the personalized product supply chain based on personalized index can be price, quality, delivery, personalization, and so on. In order to reflect the characteristics of the personalized product supply chain, we converted to the price of the negotiation through the impact of personalized index on the price. Therefore, the single-objective negotiation in this article is the negotiation between product integrator agent and consumer agent about the price.

Both the product integrator agent and the consumer agent in the personalized product supply chain have their own offer tactics. The offer tactics of the consumer agent in personalized product supply chain are as follows

p_{c}^{i} = {\begin{matrix} p_{c}^{1}, & i = 1 \\ p_{c}^{i - 1} + a_{c} (p_{s}^{i - 1} - p_{c}^{i - 1}) {(\frac{p_{c}^{i - 1}}{p_{s}^{i - 1}})}^{\frac{1}{g}}, & i \geq 2 \end{matrix}

(5)

where $p_{c}^{i}$ is the offer tactics of the consumer agent at time i; $p_{c}^{1}$ is the initial offer of the consumer agent; i (i = 1, 2, …, n) is the offer time of the consumer agent; $a_{c}$ is defined as the risk preference degree of the consumer agent, the value of which ranges from 0 to 1; if it is more than 0.5, it means that the consumer agent is risk appetite, and if it is less than 0.5, it means that the consumer agent is risk averse; g is the product’s personalized index; the greater the personalized index is, the higher the degree of personalized product is, and the smaller the personalized index is, the lower the degree of personalized product is.

The offer tactics of the product integrator agent in the personalized product supply chain are as follows

p_{s}^{i} = {\begin{matrix} p_{s}^{1}, & i = 1 \\ p_{s}^{i} = p_{s}^{i - 1} - a_{s} (p_{s}^{i - 1} - p_{c}^{i - 1}) {(p_{c}^{i - 1} / p_{s}^{i - 1})}^{g}, & i \geq 2 \end{matrix}

(6)

where $p_{s}^{i}$ is the offer tactics of the product integrator agent at time i; $p_{s}^{1}$ is the initial offer of the product integrator agent; and $a_{s}$ is defined as the risk preference degree of the product integrator agent, the value of which ranges from 0 to 1; if it is more than 0.5, it means that the product integrator agent is risk appetite and if it is less than 0.5, it means that the product integrator agent is risk averse.

The utility function of the consumer agent is monotonously decreasing, and the utility function of the product integrator agent is monotonously increasing. Here we use the linear utility function; the utility functions of the consumer agent and product integrator agent are as follows.

The price utility function of the consumer agent is as follows

U_{c}^{i} = V_{c} - \frac{p_{c}^{i} + p_{s}^{1}}{2}

(7)

The price utility function of the product integrator agent is as follows

U_{s}^{i} = \frac{p_{c}^{1} + p_{s}^{i}}{2} - V_{s}

(8)

where $V_{c}$ represents the bottom-line price of the consumer agent, that is, the highest price that the consumer agent can accept in the negotiation process, and $V_{s}$ represents the bottom-line price of the product integrator agent, which is the highest price that the product integrator agent can accept in the negotiation process.

The joint utility function is as follows

U_{cs}^{i} = a_{c} U_{c}^{i} + a_{s} U_{s}^{i}

(9)

The negotiation loss is as follows

Δ U_{c} = U_{c}^{1} - U_{c}^{i}

(10)

Δ U_{s} = U_{s}^{1} - U_{s}^{i}

(11)

Δ U = U_{cs}^{1} - U_{cs}^{i}

(12)

Negotiation tactics

We analyze the offer tactics of the product integrator agent and the consumer agent in the personalized product supply chain based on the multi-agent concept. The negotiation steps are as follows and the negotiation process is as shown in Figure 2.

Step 1. Begin to negotiate and give the initial offer

In the personalized product supply chain, the product integrator agent and the consumer agent negotiate the price of a certain personalized product, and each party has an initial quotation, namely, $p_{c}^{1}$ and $p_{s}^{1}$ . The two sides exchange the information of the initial quotation. If $p_{c}^{1} \geq p_{s}^{1}$ , the two parties will come to an agreement and the final price is $p_{s}^{1}$ , and turn to step 4. If not, turn to step 2. If one party rejects, turn to step 5.

Step 2 Negotiation, offer, and counteroffer

If $p_{c}^{i} > p_{s}^{i}$ , the two parties will come to an agreement, then turn to step 4. If not, turn to step 3. If one party rejects, turn to step 5.

Step 3. Repeating offer and counteroffer

If $p_{c}^{i} < p_{s}^{i}$ , at this time, the two parties will not come to an agreement, which requires the next round of negotiation. The two sides of the bid for the price are $p_{c}^{i + 1}$ and $p_{s}^{i + 1}$ . If $p_{c}^{i + 1} \geq p_{s}^{i + 1}$ , then the two parties will come to an agreement, and turn to step 4. If $p_{c}^{i + 1} < p_{s}^{i + 1}$ , the negotiation is not to succeed, and the two parties will not come to an agreement; this time, repeat this step until achieving a consensus price. If one party rejects, turn to step 5.

Step 4. Signing the contract and accepting the task

The negotiation mission is successful. The consumer agent will endorse the contract with the product integrator agent.

Step 5. The negotiation is to end.

Figure 2.

Flowchart of the negotiation process.

Negotiation illustrations

The negotiation example is to understand more clearly about the multi-agent single-objective negotiation model of personalized product supply chain based on personalized index, and the relationship between the two sides’ risk preference degree, product personalized degree, negotiation frequency, and joint utility. In this article, we assume that a consumer agent in a personalized product supply chain needs to purchase a personalized product that can be provided by a reliable product integrator agent. The product integrator agent and the consumer agent in the personalized product supply chain negotiate the price of the personalized product. Their bidding strategy is affected by the personalized index and is also affected by their respective risk preference, that is, $a_{s}$ and $a_{c}$ . However, the personalized index g is different from the risk preference degree $a_{s}$ and $a_{c}$ of the two sides, which leads to different prices and negotiation times, and the utility function and the joint utility function of the two sides are different. We will calculate the impact of the change of the risk preference and the personalized index on the final price, negotiation number, and negotiation loss using MATLAB. It is well known that there are lots of simulation software, such as SWARM and NetLogo. Because we are familiar with MATLAB, we chose it for simulating the negotiation process. Now, we are learning SWARM and NetLogo, and hope to use them in the future research.

According to the value of the personalized index, the influence of the personalized index on the negotiation can be divided into three categories and the exact circumstances are shown in Table 1; for ease of explanation and calculation, we set g = 0.5, 1, and 1.2, and in each case a situation is selected for detailed analysis and discussion. The initial conditions for their negotiation are as follows: $p_{c}^{1} = 600$ , $p_{s}^{1} = 800$ , $v_{c} = 800$ , and $v_{s} = 400$ , and we will analyze the effects of changes in a and g on the negotiation process.

Table 1.

The different negotiation categories.

Case 1			Case 2			Case 3
$g$	$a_{c}$	$a_{s}$	$g$	$a_{c}$	$a_{s}$	$g$	$a_{c}$	$a_{s}$
0.5	0.3	0.3, 0.5, 0.7	1	0.3	0.3, 0.5, 0.7	1.2	0.3	0.3, 0.5, 0.7
	0.5	0.3, 0.5, 0.7		0.5	0.3, 0.5, 0.7		0.5	0.3, 0.5, 0.7
	0.7	0.3, 0.5, 0.7		0.7	0.3, 0.5, 0.7		0.7	0.3, 0.5, 0.7

The first negotiation case

In this case, the personalized index g = 0.5, $a_{c} = 0.3$ , and $a_{s}$ changes from 0.3 to 0.7, according to the single-objective negotiation model steps of the personalized product supply chain based on the multi-agent concept; the negotiation process and the results are shown in Figure 3 and Table 2.

Figure 3.

The case of g = 0.5, $a_{c} = 0.3$ , and $a_{s} = 0.3, 0.5, 0.7$ .

Table 2.

Negotiation results for different risk preferences.

$g$	$(a_{c}, a_{s})$	Finalprice	Number of negotiations	$Δ U_{c}$	$Δ U_{s}$	$Δ U$
0.5	(0.3,0.3)	686.15	13	40.08	56.93	29.99
0.5	(0.3,0.5)	661.16	8	30.58	69.42	43.88
0.5	(0.3,0.7)	646.12	4	23.06	76.94	60.78

When g = 0.5, $a_{c} = 0.3$ , and $a_{s}$ is from 0.3 to 0.7, the final price is from 686.15 to 646.12 and the number of negotiations is from 13 to 4 times. The negotiation loss of $Δ U_{c}$ is from 40.08 to 23.06, the negotiation loss of $Δ U_{s}$ is from 56.93 to 76.94, and the negotiation loss of $Δ U$ is from 29.99 to 60.78.

The second negotiation case

In this case, the personalized index g = 1, $a_{c}$ = 0.3, and $a_{s}$ changes from 0.3 to 0.7, according to the single-objective negotiation model steps of the personalized product supply chain based on the multi-agent concept; the negotiation process and the results are shown in Figure 4 and Table 3.

Figure 4.

The case of g = 1, $a_{c} = 0.3$ , and $a_{s} = 0.3, 0.5, 0.7$ .

Table 3.

Negotiation results for different risk preferences.

$g$	$(a_{c}, a_{s})$	Finalprice	Number of negotiations	$Δ U_{c}$	$Δ U_{s}$	$Δ U$
1	(0.3,0.3)	700	13	50	50	30
1	(0.3,0.5)	675	8	37.5	62.5	42.5
1	(0.3,0.7)	660	4	30	70	58

When g = 1, $a_{c} = 0.3$ , and $a_{s}$ is from 0.3 to 0.7, the final price is from 700 to 660 and the number of negotiations is from 13 to 4 times. The negotiation loss of $Δ U_{c}$ is from 50 to 30, the negotiation loss of $Δ U_{s}$ is from 50 to 70, and the negotiation loss of $Δ U$ is from 30 to 58.

The third negotiation case

In this case, the personalized index g = 1.2, $a_{c} = 0.3$ , and $a_{s}$ changes from 0.3 to 0.7, according to the single-objective negotiation model steps of the personalized product supply chain based on the multi-agent concept; the negotiation process and the results are shown in Figure 5 and Table 4.

Figure 5.

The case of g = 1.2, $a_{c} = 0.3$ , and $a_{s} = 0.3, 0.5, 0.7$ .

Table 4.

Negotiation results for different risk preferences.

$g$	$(a_{c}, a_{s})$	Final price	Number of negotiations	$Δ U_{c}$	$Δ U_{s}$	$Δ U$
1.2	(0.3,0.3)	703.44	13	51.72	48.28	29.99
1.2	(0.3,0.5)	678.61	8	39.3	60.69	42.14
1.2	(0.3,0.7)	663.63	4	31.82	68.19	57.27

When g = 1.2, $a_{c} = 0.3$ , and $a_{s}$ is from 0.3 to 0.7, the final price is from 703.44 to 663.43 and the number of negotiations is from 13 to 4 times. The negotiation loss of $Δ U_{c}$ is from 51.72 to 31.82, the negotiation loss of $Δ U_{s}$ is from 48.28 to 68.19, and the negotiation loss of $Δ U$ is from 29.99 to 57.27.

Analysis and discussion

In Table 5, we summarize the three cases where the individuation index is 0.5, 1, and 1.2, and the values in brackets in the table are the transaction price, number of consultations, and the negotiation cost.

Table 5.

Negotiation results of different risk preferences and personalized index.

g	$(a_{c}, a_{s})$
g	(0.3,0.3)	(0.3,0.5)	(0.3,0.7)
0.5	(686.15, 13, 40.08, 56.93, 29.99)	(675, 8, 30.58, 69.42, 43.88)	(646.16, 4, 23.06, 76.94, 60.78)
1	(700, 13, 50, 50, 30)	(661.16, 8, 37.5, 62.5, 42.5)	(660, 4, 30, 70, 58)
1.2	(703.44, 13, 51.72, 48.28, 29.99)	(678.61, 8, 39.3, 60.69, 42.14)	(663.63.4, 31.82, 68.19, 57.27)

Different from related works, we mainly discuss the influence of different risk preferences of agent and personalized index on the negotiation process:

In this article, the offer of the seller is an increasing function of the personalized index g and is a convex function. When the other conditions are constant, the final price is increasing with the increase of g. On the basis of the basic functions of any product, adding some functions, or choosing the best supplier, will increase the cost of the personalized product, and of course the price of the personalized product will also increase. The change of the personalized index g has no obvious effect on the number of negotiations. The greater the g, the greater the negotiation loss of the customer and the more the final price deviating from the customer’s initial offer, which means that the customer will pay the higher price to buy the personalized product. The greater the g, the smaller the negotiation loss of the product integrator, which means that the final price is closer to the initial offer of the product integrator.

It can be seen from the table that the change of the risk preference has an obvious effect on the final price, negotiation number, and negotiation loss. When the parties are not preferred to risk, the two sides are very cautious to offer and the price change is very small, which can reach agreement for more rounds of negotiation. The more the risk preference, the greater the negotiation loss and the less the negotiation numbers. The final price is more conducive to the party who does not prefer the risk.

Conclusion

The effective operation of personalized product supply chain will increase the customer satisfaction. Negotiation will increase the operational efficiency of personalized product supply chain management. In this article, we expatiate the personalized index and present the calculation method. We set the multi-agent negotiation model and tactics based on personalized index. And finally the illustration is given for verifying the validity of the model. This will help to more clearly understand and learn the multi-agent negotiation essence of personalized product supply chain and to improve the operational efficiency of personalized product supply chain.

We mainly consider the negotiation of two parties in this article. And in the future research we will conduct negotiation among the three parties on a single objective or multiple objectives. In the actual operation of personalized product supply chain, the negotiation about three parties is widespread. With the improvement of consumption level and application of Internet technology, manufacturing enterprise personalized product supply chain has become a trend, the negotiation is not the same between the personalized product supply chain multi-agent system and traditional manufacturing enterprise supply chain multi-agent system, and this problem is worth studying. We would like to cooperate with researchers in this field to carry out such research together.

Footnotes

Handling Editor: Peter Nielsen

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Natural Science Foundation of China (Nos 71702174 and U1404704). And this work was supported by the Soft Science Program Research Project of Henan (No. 182400410281).

ORCID iD

Changhui Yang

References

Hendrik

Nadine

Interconnecting product and process information to enable personalized production. Proc CIRP 2016; 52: 186–191.

Zhao

Zhang

et al . Research on comprehensive evaluation of rural residents consumption level based on factor analysis model. J Inform Technol Manage Sci 2017; 2: 27–31.

Zhao

Song

Chang

et al . Research on coal production logistics efficiency of evaluation index system based on DEA. J Inform Technol Supply Chain Manage 2017; 2: 58–63.

Guritno

Tanuputri

Evaluation method of weapon equipment maintenance capability based on job completed degree. J Inform Technol Manage Sci 2017; 2: 62–66.

Sheng

Cheng

Optimality analysis of port coal logistics collection and distribution system. J Inform Technol Supply Chain Manage 2017; 2: 7–12.

Mintzberg

Lampel

Customizing customization. Sloan Manage Rev 1996; 38: 21–30.

Gilmore

Pine

BJN

. The four faces of mass customization. Harvard Bus Rev 1997; 72: 91–101.

Maohua

Research on service supply chain operation model based on personalized demand. PhD Thesis, Beijing Jiaotong University, Beijing, China, 2012.

Zhijun

Research on personalized product requirements management and rapid customization technology. PhD Thesis, Shandong University, Jinan, China, 2012.

10.

Zhihua

Research on logistics service based on personalized demand of enterprises. PhD Thesis, Huazhong Normal University, Wuhan, China, 2012.

11.

Xiaodong

Xuegang

On product attribute configuration for multiple personalized demands. Ind Eng J 2013; 16: 54–60.

12.

Sun

Y-Y

Fui

Importance determining method of personalized product attributes based on Kano-QFD integration model. Comput Integr Manuf 2014; 20: 2697–2704.

13.

Peppers

Rogers

Dorf

Is your company ready for one-to-one marketing?

Harvard Bus Rev 1999; 77: 151–160.

14.

Wind

Rangaswamy

Customerization: the next revolution in mass customization. J Interact Mark 2001; 15: 13–32.

15.

Siddique

Boddu

. A CAD template approach to web-based customer centric product design. J Comput Inf Sci Eng 2005: 54381–54387.

16.

Anthony

Elli

Dynamic generation of personalized product bundles in enterprise networks. Lect Notes Comput Sc 2013; 8186: 208–217.

17.

Brenner

Hummel

A seamless convergence of the digital and physical factory aiming in personalized product emergence process (PPEP) for smart products within ESB Logistics Learning Factory at Reutlingen University. Proc CIRP 2016; 54: 227–232.

18.

Chikhaoui

Lotfi

Research on model of solving container stowage planning problem. J Inform Technol Supply Chain Manage 2017; 2: 71–77.

19.

Lane

Mcfadzean

AG.

Distributed problem solving and real-time mechanisms in robot architectures. Eng Appl Artif Intel 1994; 7: 105–117.

20.

Yang

Sun

. Research on negotiation of manufacturing enterprise supply chain based on multi-agent. J Internet Technol, in press.

21.

Yang

et al . Negotiation model and tactics of manufacturing enterprise supply chain based on multi-agent, Adv Mech Eng 2018; 10: 1–8.

22.

Shoham

Agent-oriented programming

Artif Intell 1993; 60: 51–92.

23.

Wooldridge

Jennings

NR.

Intelligent agents: theory and practice. Knowl Eng Rev 1995; 10: 115–152.

24.

Wooldridge

Agent-based software engineering. IEEE Proc Softw Eng 1997; 144: 26–37.

25.

Franklin

Graesser

. Is it an agent, or just a program? A taxonomy for autonomous agent. In: Intelligent Agents III ATAL 1996 (eds. Müller

Wooldridge

Jennings

), Lecture Notes in Computer Science (Lecture Notes in Artificial Intelligence), vol 1193, pp. 21–35. Berlin, Heidelberg: Springer.

26.

Maturana

Norrie

DH.

Multi-agent mediator architecture for distributed manufacturing. J Intell Manuf 1996; 7: 257–270.

27.

Labrou

Finin

. A proposal for a new KQML specification, DARPA knowledge sharing initiative. Exterfaces Working Group working paper, 3 February 1997. https://www.csee.umbc.edu/csee/research/KQML/papers/kqml97.pdf

28.

FIPA. FIPA contract net interaction protocol specification, http://www.fipa.org/specs/fipa00029

29.

Zhang

Liu

Duan

Research on buy back contract application in supply chain coordination. J Inform Technol Supply Chain Manage 2017; 2: 93–97.

30.

Durfee

Lesser

Corkill

Trends in cooperative distributed problem solving. IEEE T Knowl Data En 1989; 1: 63–83.

31.

Neumann

Morgenstern

The theory of games and economic behavior. New York: Princeton University Press, 1994.

32.

Faratin

Sierra

Jennings

NR.

Negotiation decision functions for autonomous agent. Robot Auton Syst 1998; 24: 159–182.

33.

Jennings

Norman

Faratin

et al . Autonomous agents for business process management. Appl Artif Intell 2000; 14: 145–189.

34.

Kraus

Automated negotiation and decision making in multi-agent environments. Lect Notes Comput Sc 2001; 2086: 150–172.

35.

Kraus

Wilkenfeld

Zlotkin

Multi-agent negotiation under time constraints. Artif Intell 1995; 75: 297–345.

36.

Fatima

Wooldridge

Multi-issue negotiation under time constraints. In: Proceedings of the 1st international joint conference on autonomous agents and multi-agent systems, Bologna, Italy, 15–19 July 2002, pp. 143–150. New York, NY, USA: ACM.

37.

Fatima

Wooldridge

Jennings

NR.

An agenda-based framework for multi-issue negotiation. Artif Intell 2004; 152: 1–45.

38.

Park Yang

. An efficient multilateral negotiation system for pervasive computing environments. Eng Appl Artif Intel 2008; 21: 633–643.

39.

Sheng

A multi-agent model for the reasoning of uncertainty information in supply chains. Int J Prod Res 2011; 49: 5737–5753.

40.

Huang

C-J

Liao

L-M.

A multi-agent-based negotiation approach for parallel machine scheduling with multi-objectives in an electro-etching process. Int J Prod Res 2012; 50: 5719–5733.

41.

Sunil

Mittal

Abhinav

A multi-agent system for decentralized multi-project scheduling with resource transfers. Int J Prod Econ 2013; 146: 646–661.

42.

Chen

Weiss

An efficient automated negotiation strategy for complex environments. Eng Appl Artif Intel 2013; 26: 2613–2623.

43.

Hernández

Mula

Poler

et al . Collaborative planning in multi-tier supply chains supported by a negotiation-based mechanism and multi-agent system. Group Decis Negot 2014; 23: 235–269.

44.

Khaled

Besoa

Analysis of the performance of supply chains configurations using multi-agent systems. INT J Logist-Res App 2014; 17: 441–458.

45.

Mansour

Kowalczyk

Coordinating the bidding strategy in multi-issue multi-object negotiation with single and multiple providers. IEEE T Cybernetics 2015; 45: 2261–2272.

46.

Manupati

Putnik

Tiwari

et al . Integration of process planning and scheduling using mobile-agent based approach in a networked manufacturing environment. Comput Ind Eng 2016; 94: 63–73.

47.

Ilany

Gal

Algorithm selection in bilateral negotiation. Auton Agent Multi-Ag 2016; 30: 697–723.