Seller Agents Research Articles

MODELING AND SIMULATION OF SEQUENTIAL AUCTIONS: PRICING AND CALIBRATION ALGORITHMS

We consider sequential auctions wherein seller and bidder agents need to price goods on sale at the 'right' market price. We propose algorithms based on a binomial model for both the seller and buyer. Then, we consider the problem of calibrating pricing models to market data. To this end, we studied a stochastic volatility model used for option pricing, derived, and analyzed Monte Carlo estimators for computing the gradient of a certain payoff function using Finite Differencing and Algorithmic Differentiation. We then assessed the accuracy and efficiency of both methods as well as their impacts into the optimization algorithm. Numerical results are presented and discussed. This work can benefit those engaged in electronic trading or investors in financial products with the need for fast and more precise predictions of future market data.

Reserve price strategy for seller agent in multiple simultaneous auctions

Internet auction is popular due to the flexibility and convenience that it offers to consumers. In online auctions, sellers are confronted with the dilemma of deciding the best reserve price for the items to be auctioned off. In an auction site such as eBay, one can always find the same item being sold by multiple sellers in different auctions. Determining this reserve price is not a straightforward decision process due to the complexity and vagueness of the online auction environment. Setting the reserve price too high may result in no sale whilst setting the reserve price too low may result in a sale with low profit. The main focus of this paper is to analyze the performance of the selling agents with varying pricing strategy when offering item for auctions. The strategy could be categorized as low, intermediate, high, and random based on its sellers' types. Our study showed that the use of these strategies produced significant drawbacks with negative impacts towards the overall selling upshots. To counteract these shortcomings, we develop an autonomous seller agent using a heuristic decision making framework. To derive the best reserve price, several constraints are considered including the number of competitors, the number of bidders, the auction length, and the profit that the seller desires. This paper presents the design, implementation and evaluation of our selling algorithm that a seller agent can use for auctioning homogeneous goods among multiple overlapping English auctions. In this work, we modeled our market simulation for a single unit auction using an independent private value framework with dynamic participation entry.

Autonomous Seller Agent for Multiple Simultaneous English Auctions

The growth of online auction is due to the flexibility and convenience that it offers to consumers. In the context of online auction, deriving the best reserve price can be associated to the seller’s optimization problem. Determining this reserve price is not straightforward due to the dynamic and unpredictable nature of the auction environment. Setting the price too high will lead to the possibility of no sale outcome. Putting the price too low may produce a sale with less profit due to its lower selling price. The authors propose a strategy to derive the best reserve price based on several selling constraints such as the number of competitors (sellers), the number of bidders, the auction duration, and the profit the seller desired when offering an item to be auctioned. However, to obtain the best performance, the strategy must be tuned to the prevailing auction environment where the agent is situated. This paper describes the seller agent’s performance under varying auction environments. The purpose of the experimental evaluation is to assess the ability of the agent to identify its environments accurately to enable it to come up with the best reserve price.

Personality for buyer and seller agents in electronic marketplace based on reputation and reinforcement learning

In this paper, we propose a marketplace model which is based on personality, reputation and reinforcement learning algorithms for buying and selling agents. We use two personality traits for seller agents: stingy and conscientiousness. In this marketplace, sellers with low score of stingy earn more benefits in comparison with high stingy sellers. Also, conscientious seller agents gain more reputation relative to conscienceless seller agents. In addition, we use three personality traits for buyer agents: stingy, openness and agreeableness. Buyer agents with high score of openness and low score of stingy purchase more new goods and more expensive goods relative to buyers with low score of openness and high score of stingy. Buyer agent’s seller with high score of agreeableness would be less cheated in the marketplace. Also, buyer agents apply reinforcement learning to evaluate the reputation of seller agents and then focus their trading on reputable sellers. On the other hand, the personality of seller agents affects them to consider discount for buyer agents. In addition, seller agents apply reinforcement learning to establish a model of reputation of buyer agents. The results show that selling/buying agents that model the reputation of buying/selling agents obtain more satisfaction rather than selling/buying agents who only use the reinforcement learning. Key words: Agent, marketplace, personality, reputation, reinforcement learning.

A multi-agent recommender system for supporting device adaptivity in e-Commerce

Traditional recommender systems for e-Commerce support the customers' activities providing them with useful suggestions about available products in Web stores. To this purpose, in an agent-based context, each customer is often associated with a customer agent that interacts with the site agent associated with the visited e-Commerce Web site. In presence of a high number of interactions between customers and Web sites, the generation of recommendations can be a heavy task for both these agents. Moreover, customers can navigate on the Web by using different devices having different characteristics that may influence customer's preferences. In this paper we propose a new multi-agent system, called ARSEC, where each device exploited by a customer is associated with a device agent that autonomously monitors his/her behaviour. Furthermore, each customer is associated with a customer agent that collects in a global profile the information provided by his/her device agents and each e-Commerce Web site is associated with a seller agent. Based on the similarity existing among the global profiles the customers are partitioned in clusters, each one managed by a counsellor agent. Recommendations are generated in ARSEC as result of the collaboration between the seller agent and some counsellor agents associated with the customer. The usage of the device agents leads to generating recommendations taking into account the device currently used, while the fully decentralized architecture introduces a strong reduction of the time costs. Some experimental results are presented to show the significant advantages obtained by ARSEC in terms of recommendation effectiveness with respect to other well-known agent-based recommenders.

Multi-Agent Negotiation Paradigm for Agent Selection in B2C E-Commerce

Multi agent system (MAS) model has been extensively used in the different tasks of E-Commerce such as customer relation management (CRM), negotiation and brokering. The objective of this paper is to evaluate a seller agent’s various cognitive parameters like capability, trust, and desire. After selecting a best seller agent from ordering queue, it applies negotiation strategies to find the most profitable proposal for both buyer and seller. This mechanism belongs to a semi cooperative negotiation type, and selecting a seller and buyer agent pair using mental and cognitive parameters. This work provides a logical cognitive model, logical negotiation model between buyer agent and selected seller agent.

Enhanced Reputation Model with Forgiveness for E-Business Agents

Trust is a very important quality attribute of an e-service. In particular, the increasing complexity of the e-business environment requires the development of new computational models of trust and reputation for e-business agents. In this paper, the authors introduce a new reputation model for agents engaged in e-business transactions. The model enhances classic reputation models by the addition of forgiveness factor and the use of new sources of reputation information based on agents groups. The paper proposes an improvement of this model by employing the recent con-resistance concept. Finally, the authors show how the model can be used in an agent-based market environment where trusted buyer and seller agents meet, negotiate, and transact multi-issue e-business contracts. The system was implemented using JADE multi-agent platform and initially evaluated on a sample set of scenarios. The paper introduces the design and implementation of the agent-based system together with the experimental scenarios and results.

Automatic price negotiation on the web: An agent-based web application using fuzzy expert system

A traditional internet auction is restricted by the limitation of time. It is necessary to conduct an internet auction in a certain time period. The final trading price is determined until this certain period ends. This study improves this situation by removing the time limitation. Based on the fuzzy inference theory, this paper proposes an agent-based price negotiation system for on-line auctions. Mainly, three agents are used in the study: a seller agent, a buyer agent, and a mediator agent. The proposed system provides an easy-to-use environment and good customizability for users (buyers or sellers) to customize their price negotiation strategies using user-defined fuzzy rules. The final negotiated price is immediately determined after the buyer sends his bids to the proposed system. This study develops a Java-based computer package to implement the price negotiation system where Model-View-Controller (MVC) design pattern is employed in design of the package. Unified Modeling Language (UML) is also utilized to describe the structures and behaviors of the package. To validate the proposed system, this study built an on-line auction website with the proposed price negotiation mechanism for internet users to buy or sell their merchandises. An evaluation was finally conducted to investigate the users’ satisfaction with the proposed system.Briefly, the proposed system is featured by: (1) instantly getting negotiated price without waiting; (2) conducting price negotiation at any time; (3) determining strategy rules easily, and (4) using customizable negotiation strategies defined by users.

Multi-Agent Negotiation in B2C E-Commerce Based on Data Mining Methods

The Multi agent system (MAS) model has been extensively used in the different tasks of e-commerce like customer relation management (CRM), negotiation and brokering. For the success of CRM, it is important to target the most profitable customers of a company. This paper presents a multi-attribute negotiation approach for negotiation between buyer and seller agents. The communication model and the algorithms for various actions involved in the negotiation process is described. The paper also proposes a multi-attribute based utility model, based on price, response-time, and quality. In support of this approach, a prototype system providing negotiation between buyer agents and seller agents is presented.

Customer Orientation Based Multi-Agent Negotiation for B2C e-Commerce

The Multi Agent System (MAS) model has been extensively used in different e-commerce tasks. Customer orientation is a crucial component of organizational culture and attention to information about customers’ needs should be considered a firm’s most basic value. In this paper, the authors evaluate a seller agent’s various cognitive parameters, such as capability, trust, and desire. The authors also examine customer orientation value for a pair of buyer and seller agents, such as selecting a seller and buyer agent pair using mental and cognitive parameters, customer orientation value for their mutual benefits, and the competitive negotiation factor. The authors provide a deterministic customer orientation computational model, a competitive negotiation computational model between buyer agent and seller agent, and a logical computational model for selection of the best buyer-seller agent pair. With the help of GUI intermediate, final results are shown.

Openness, Conscientiousness and Stingy for Buyerand Seller Agents in Electronic Marketplace

In this paper we propose a model based on personality for buyer and seller agents in agent-based electronic marketplaces. The personality of buyer agents affects their behavior in market. Buyer agents use their own personality to evaluate the value of seller agents' bids. Also buyer agents apply reinforcement learning to evaluate the reputation of seller agents and then focus their trading on reputable sellers. On the other hand, the personality of seller agents affects them to consider discount for buyer agents. In addition, seller agents apply reinforcement learning to model the reputation of buyer agents. We have implemented this model with aglet which is a java based environment for building agents. Our results show that sellers with low score of stingy earn more benefits in comparison with high stingy sellers. Also, conscientious seller agents gain more reputation relative to conscienceless seller agents. On the other hand, buyer agents with high score of openness and low score of stingy purchase more new goods and more expensive goods relative to buyers with low score of openness and high score of stingy.

Analysis of single and networked auctions

Web-based computerized auctions are increasingly present in the Internet. We can imagine that in the future this trend will actually be extended to situations where virtual buyer and seller agents will conduct automated transactions across the network, and that large sectors of the economy may be strucured in this manner. The purpose of this article is to model automated bidders and sellers which interact through a network. We model the bidding process as a random arrival process while the price attained by a good is modeled as a discrete random variable. We obtain analytical solutions allowing us to compute the income from a single auction, or the income per unit time from a repeated sequence of auctions. A variety of single-auction models are studied, including English and Vickrey auctions, and the income per unit time is derived as a function of other parameters, including the rate of arrival of bids, the seller's decision time, the value of the good, and the “rest time” of the seller between successive auctions. We illustrate the results via numerical examples. We also introduce a model for networked auctions where bidders can circulate among a set of interconnected auctions which we call the Mobile Bidder Model (MBM). We obtain an analytical solution for the MBM under the assumption,which we call the “active bidders assumption,” that activities that are internal to an auction (bids and sales) are much more frequent than changes that occur in the number of bidders at each auction.

Modelling Bidders in Sequential Automated Auctions

Auctions are mechanisms that formalize the rules with which automated trading schemes can be conducted, and in this paper we model the interaction of bidder and seller agents in sequential computerized auctions. We study the outcome of strategies that a designated ‘special bidder’ (SB) may follow in the presence of a collection of other bidders in an English auction, under the assumption that the SB can make bids based on its observation of the ongoing auction as a collective system. In our model, bidding and sale events are continuous time random processes with discrete state-space, where the state-space represents the current value of the most recent bid. We obtain analytical solutions that allow the evaluation of measures of interest to the SB such as the probability of winning, the savings with respect to the maximum payable price in the event of a win and the expected waiting time to win. We examine the effects of the SB's time to bid, and study how its decisions may be selected so as to optimize the SB's measures of interest.

The profit implications of altruistic versus egoistic orientations for business-to-business exchanges

This study significantly expands upon previous research by Hill and Watkins [Hill, Ronald Paul and Watkins, Alison, (2007), “A Simulation of Moral Behavior within Marketing Exchange Relationships,” Journal of the Academy of Marketing Science, 35 (2), 417–429] involving business-to-business exchanges through the use of a more sophisticated simulation and a different theoretical orientation. Profitability implications for sellers and firms in the context of information sharing and dynamic firm adaptation are explored using q-learning evolutionary models and embedded artificial trading agents in a competitive environment. This method allows buyer agents to react to complex and evolving circumstances based on historical information about seller agents. The results suggest that sellers with more cooperative strategies are more profitable in the long run, especially when firms employ multiple agents.

A heuristic personality-based bilateral multi-issue bargaining model in electronic commerce

Our everyday lives and specially our commercial transactions involve complex negotiations that incorporate decision-making in a multi-issue setting under utility constraints. Negotiation as a key stage in all commercial transactions has been proliferated by applying decision support facilities that AI techniques provide. Recently, Distributed Artificial Intelligence techniques have been evolved towards multi-agent systems (MASs) where each agent is an intelligent system that solves a specific problem. Incorporating MAS into e-commerce negotiation and bargaining has brought even more potential improvement in efficiency and effectiveness of business systems by automating several of the most time consuming and repetitive stages of the buying process. In bargaining, participants with opposing interests communicate and try to find mutually beneficial agreements by exchanging compromising proposals. However, recent studies on commercial bargaining and negotiation in MASs lack a personality model. Indeed, adding personality to intelligent agents makes them more human-like and increases their flexibility. We investigate the role of personality behaviors of participants in multi-criteria bilateral bargaining in a single-good e-marketplace, where both parties are OCEAN agents based on the five-factor (Openness, Conscientiousness, Extraversion, Agreeableness, and Negative emotions) model of personality. We do not aim to determine strategies that humans should use in negotiation, but to present a more human-like model to enhance the realism of rational bargaining behavior in MASs. First, this study presents a computational approach based on a heuristic bargaining protocol and a personality model, and second, considers the issue of what personality traits and behaviors should be investigated in relation to automated negotiations. We show the results obtained via the simulation on artificial stereotypes. The results suggest and model compound personality style behaviors appropriate to gain the best overall utility in the role of buyer and seller agents and with regard to social welfare and market activeness. This personality-based approach can be used as a predictive or descriptive model of human behavior to adopt in appropriate situations in many areas involving negotiation and bargaining (e.g., commerce, business, politics, military, etc.) for conflict prevention and resolution. This model can be applied as a testbed for comparing personality models against each other based on human data in different negotiation domains.

Evaluating the trustworthiness of advice about seller agents in e-marketplaces: A personalized approach

In this paper, we present a model for evaluating the trustworthiness of advice about seller agents in electronic marketplaces. In particular, we propose a novel personalized approach for effectively handling unfair ratings of sellers provided to buyer agents from other buyers (called advisors). Our approach offers flexibility for buyers to weight their value for private and public knowledge about advisors. A personalized approach is proposed as well for buyers to model the trustworthiness of sellers, based on the advice provided. Experimental results demonstrate that our approach can effectively model trustworthiness for both advisors and sellers, even when there are large numbers of unfair ratings.

Anegsys: An automated negotiation based recommender system for local e-marketplaces

Local e-marketplaces are local online e-commerce platforms deployed by product and service providers and accessed by local customers via mobile devices. In this scenario, customers need to gather information about available offers from the different providers in the area, in order to select the most suitable for their needs and preferences. We present ANEGSYS, an agent-based recommender system for product acquisition which uses automatic bilateral negotiations to generate purchase pre-agreements among buyer and seller agents. This greatly enhances the search for solutions which maximize both buyer and seller utilities.

An Electronic Marketplace Based on Reputation and Learning

In this paper, we propose a market model which is based on reputation and reinforcement learning algorithms for buying and selling agents. Three important factors: quality, price and delivery-time are considered in the model. We take into account the fact that buying agents can have different priorities on quality, price and delivery-time of their goods and selling agents adjust their bids according to buying agents preferences. Also we have assumed that multiple selling agents may offer the same goods with different qualities, prices and delivery-times. In our model, selling agents learn to maximize their expected profits by using reinforcement learning to adjust product quality, price and delivery-time. Also each selling agent models the reputation of buying agents based on their profits for that seller and uses this reputation to consider discount for reputable buying agents. Buying agents learn to model the reputation of selling agents based on different features of goods: reputation on quality, reputation on price and reputation on delivery-time to avoid interaction with disreputable selling agents. The model has been implemented with Aglet and tested in a large-sized marketplace. The results show that selling/buying agents that model the reputation of buying/selling agents obtain more satisfaction rather than selling/buying agents who only use the reinforcement learning.

An adaptive attitude bidding strategy for agents in continuous double auctions

A continuous double auction (CDA) is an efficient market institution for real-world trading of commodities and electronic marketplaces. In this paper, we present the design and analysis of a new bidding strategy for buyer and seller agents participating in agent-based CDAs. The strategy employs heuristic rules and a reasoning mechanism based on a two-level adaptive bid-determination method, including short-term and long-term attitudes. Agents adopting the strategy dynamically adjust their behaviors in response to the changes of the supply and demand relationships in the market. Experimental results show that agents adopting the strategy outperform agents using other strategies reported in the literature.

Modeling Agent Auctions in a Supply Chain Environment

Agent-based auction technology has revolutionized auction trading in the Supply Chain environment by reducing the cost of transactions, and by increasing the satisfaction factor in matching requirements of seller and buyer agents. In this article, we have considered methods of matching quantities of buyer and seller agents by cooperation, with a priority on the buyer’s requirements. The article discusses the architecture of the agent and the agent community when there is cooperative matching of volume. We present a Dynamic Programming algorithm to describe the agent’s decision process, and heuristic algorithms as the practical solution methodology. The results of a simple experiment show the improvement achieved by cooperation.