Determining the Optimal Reporting Strategy in Competitive E-Marketplaces Zeinab Noorian 1 Jie Zhang 2 Michael Fleming 1 Stephen Marsh 3 1 Faculty of Computer Science, University of New Brunswick, Canada 2 School of Computer Engineering, Nanyang Technological University, Singapore 3 Communications Research Centre, Canada Email: z.noorian@unb.ca Abstract—In a reputation system for multiagent based elec- tronic marketplaces where the number of high quality products provided by good selling agents is unlimited, buying agents often share seller information without the need to consider possible utility loss. However, when those good sellers have limited inventory, buyers may have to be concerned about the possibility of losing the opportunity to do business with the good sellers if the buyers provide truthful information about sellers, due to the competition from other buyers. In this paper, we propose an adaptive mechanism built on a game theoretic basis for buyers to determine their optimal reputation reporting strategy, by modeling both the competency and willingness of other buyers in reporting seller reputation and strategically choosing reporting behaviours that maximize their utility according to the modeling results. The results of the experiments carried out in a simulated competitive e- marketplace confirm that our proposed mechanism leads to better utility for buyers in such an environment. Keywords-Trust and Reputation, Game Theory, Behavioural Modeling, Seller Selection, Competitive E-Marketplaces I. I NTRODUCTION In open multiagent based e-marketplaces, some selling agents may be malicious and may not deliver products with the same quality as what they promised. Thus, buying agents need a means to assess the quality of different sellers offering a particular product and select the most profitable seller who best meets the buyers’ requirements. Reputation systems [1], [2], [3], [4] are a particularly effective approach for buyers to model sellers’ reputation (representing their quality) based on the reporting of seller information provided by other buyers (also called advisers). These systems often assume that sellers have infinite (or very large) inventory and the number of high quality products provided by good sellers is unlimited. Also, a successful business transaction of one buyer would not result in a loss for other buyers. Thus, in such environments, buyers can report seller reputation in- formation according to their own endogenous characteristics without considering the possible utility loss as the result of their truthful reporting and competition from others. However, in certain e-marketplaces, good sellers may have limited inventory. One example is the hotel booking system for a famous tourism area during a peak season since booking a satisfactory hotel is often difficult. Similar marketplaces also include second-hand markets where some used and workable goods (e.g. second-hand textbooks) are often in short supply due to lower prices. Also, different buyers may aim for the same kind of high quality products. These buyers compete to discover the high quality sellers who will maximize their utility in order to conduct business transactions with these sellers before their stock runs out. In these competitive e-marketplaces, a buyer may have to be concerned about the possibility of losing the oppor- tunity to do business with good sellers if providing truthful reputation information about sellers. To be more specific, after some successful transactions with a seller, if the buyer provides truthful (positive) feedback about the good seller, the buyer may lose the chance to do business with the seller in the future, due to the limited inventory the seller has and the fact that other buyers will also do business with this good seller. If the transactions are unsuccessful, reporting truthful (negative) feedback may cause the buyer to lose the chance to do business with other good sellers because other buyers will not do business with the bad seller but with the other good sellers, after taking the buyer’s advice. In this sense, it is better for buyers not to truthfully reveal seller reputation. On the other hand, buyers are also motivated to participate in information exchange because truthful sharing of seller reputation allows for faster discovery of high quality sellers. It is thus not trivial to determine an optimal reporting strategy for buyers that maximizes their utility in competitive e-marketplaces, and this issue has not yet been well addressed in the literature. Based on the above discussions, we intuit that other buyers (advisers) may not behave as expected in competitive e- marketplaces. That is, the reporting behaviour of the advisers may not only be dependent on their endogenous characteris- tics (i.e. competency), because competent advisers may not always be willing to cooperate with buyers by reporting truthful reputation information about sellers. Thus, when choosing their reporting behaviour, buyers should carefully examine the trustworthiness (quality) of advisers in reporting seller reputation information, by also taking into account the willingness of advisers. In additions, buyers should have in mind both possible competition and provision of valuable seller information from advisers. Therefore, in this paper,