Valentin Robu

Automated Multi-Attribute Negotiation with Efficient Use of Incomplete Preference Information

This paper presents a model for integrative, one-to-one negotiation in which the values across multiple attributes are negotiated simultaneously. We model a mechanism in which agents are able to use any amount of incomplete preference information revealed by the negotiation partner in order to improve the efficiency of the reached agreements. Moreover, we show that the outcome of such a negotiation can be further improved by incorporating a guessing heuristic, by which an agent uses the history of the opponents bids to predict his preferences. Experimental evaluation shows that the combination of these two strategies leads to agreement points close to or on the Pareto-efficient frontier. The main original contribution of this paper is that it shows that it is possible for parties in a cooperative negotiation to reveal only a limited amount of preference information to each other, but still obtain significant joint gains in the outcome.

Decommitment in a competitive multi-agent transportation setting

Decommitment is the action of foregoing of a contract for another (superior) offer. It has been analytically shown that, using decommitment, agents can reach higher utility levels in case of negotiations with uncertainty about future opportunities. We study the decommitment concept for the novel setting of a large-scale logistics setting with multiple, competing companies. Orders for transportation of loads are acquired by agents of the (competing) companies by bidding in online auctions. We find significant increases in profit when the agents can decommit and postpone the transportation of a load to a more suitable time. Furthermore, we analyze the circumstances for which decommitment has a positive impact if agents are capable of handling multiple contracts simultaneously. Lastly, we present a demonstrator of the developed model in the form of a Java Applet.

Simulation and Visualization of a Market-Based Model for Logistics Management in Transportation

Distributed logistics and transportation is an important and emerging area of application for multi-agent systems, which has recently attracted a lot of research interest. In previous research ([1], [2]) we have proposed and developed novel techniques to deal with some of the challenges and problems in this application domain. In this paper we describe the software system which was built to visualize and demonstrate our multi-agent model.

A System for Analysis of Multi-Issue Negotiation

This paper presents a System for Analysis of Multi-Issue Negotiation (SAMIN). The agents in this system conduct one-to-one negotiations, in which the values across multiple issues are negotiated on simultaneously. It is demonstrated how the system supports both automated negotiation (i.e., conducted by a software agent) and human negotiation (where humans specify their bids). To analyse such negotiation processes, the user can enter any formal property deemed useful into the system and use the system to automatically check this property in given negotiation traces. Furthermore, it is shown how, compared to fully closed negotiation, the efficiency of the reached agreements may be improved, either by using incomplete preference information revealed by the negotiation partner or by incorporating a heuristic, through which an agent uses the history of the opponent’s bids in order to guess his preferences.

Designing Risk-Averse Bidding Strategies in Sequential Auctions for Transportation Orders

Designing efficient bidding strategies for agents participating in multiple, sequential auctions remains an important challenge for researchers in agent-mediated electronic markets. The problem is particularly hard if the bidding agents have complementary (i.e. super-additive) utilities for the items being auctioned, such as is often the case in distributed transportation logistics. This paper studies the effect that a bidding agent’s attitude towards taking risks plays in her optimal, decision-theoretic bidding strategy. We model the sequential bidding decision process as an MDP and we analyze, for a category of expectations of future price distributions, the effect that a bidder’s risk aversion profile has on her decision-theoretic optimal bidding policy. Next, we simulate the above strategies, and we study the effect that an agent’s risk aversion has on the chances of winning the desired items, as well as on the market efficiency and expected seller revenue. The paper extends the results presented in our previous work (reported in [1]), not only by providing additional details regarding the analytical part, but also by considering a more complex and realistic market setting for the simulations. This simulation setting is based on a real transportation logistics scenario [2]), in which bidders have to choose between several combinations (bundles) of orders that can be contracted for transportation.

The Complex Dynamics of Sponsored Search Advertising Markets

This paper provides a comprehensive study of the structure and dynamics of online advertising markets, mostly based on techniques from the emergent discipline of complex systems analysis. First, we look at how the display rank of a URL link influences its click frequency, for both sponsored search and organic search. Second, we study the market structure that emerges from these queries, especially the market share distribution of different advertisers. We show that the sponsored search market is highly concentrated, with less than 5% of all advertisers receiving over 2/3 of the clicks in the market. Furthermore, we show that both the number of ad impressions and the number of clicks follow power law distributions of approximately the same coefficient. However, we find this result does not hold when studying the same distribution of clicks per rank position, which shows considerable variance, most likely due to the way advertisers divide their budget on different keywords. Finally, we turn our attention to how such sponsored search data could be used to provide decision support tools for bidding for combinations of keywords. We provide a method to visualize keywords of interest in graphical form, as well as a method to partition these graphs to obtain desirable subsets of search terms.