Towards automated procurement via agent-aware negotiation support ∗ A. Giovannucci, J. A. Rodr´ ıguez-Aguilar IIIA-CSIC, Campus UAB 08193 Bellaterra, Barcelona, Spain {andrea,jar}@iiia.csic.es A. Reyes, F. X. Noria, J. Cerquides Intelligent Software Components, S. A. 08190 Sant Cugat del Vall` es, Barcelona, Spain {toni,fxn,cerquide}@isoco.com Abstract Negotiation events in industrial procurement involv- ing multiple, highly customisable goods pose serious chal- lenges to buying agents when trying to determine the best set of providing agents’ offers. Typically, a buy- ing agent’s decision involves a large variety of con- straints that may involve attributes of a very same item as well as attributes of multiple items. In this paper we de- scribe iBundler, an agent-aware negotiation service to solve the winner determination problem considering buy- ers’ and providers’ constraints and preferences. 1. Introduction Consider the problem faced by a buying agent when ne- gotiating with providing agents. In a negotiation event in- volving multiple, highly customisable goods, buying agents need to express relations and constraints between attributes of different items. Moreover, it is common practice to buy different quantities of the very same product from differ- ent providing agents, either for safety reasons or because offer aggregation is needed to cope with high-volume de- mands. This introduces the need to express business con- straints on providing agents and the contracts they may have assigned. Not forgetting the provider side, providing agents may also wish to impose constraints or conditions over their offers. These may only be valid if certain con- figurable attributes (e.g. quantity, delivery days) fall within some intervals, or assembly and packing constraints need to be considered. Once a buying agent collects all offers, he is faced with the burden of determining the winning of- fers. The problem is essentially an extension of the combi- natorial auction (CA) problem, which can be proved to be NP-complete[14]. It would be desirable to relieve buying agents from solving such a problem. In this paper we have tried to make headway in this direction by fully describing * Partially supported by project Web-i(2) (TIC-2003-08763-C02-01). iBundler (extended the work in [13]),an agent-aware deci- sion support service acting as a combinatorial negotiation solver (solving the winner determination problem) for both multi-item, multi-unit negotiations and auctions. Thus, the service can be employed by both buying agents and auc- tioneers in combinatorial negotiations and combinatorial re- verse auctions[15] respectively. Furthermore, it extends cur- rent CA models by accommodating both operational con- straints and attribute-value constraints. At this aim, new on- tological issues needed to be considered in order to em- power the expressiveness offered by negotiation objects and offers to incorporate buyers’ and providers’ business con- straints. Therefore, our approach required the extension of state-of-the-art ontologies for automated negotiation. To the best of our knowledge, iBundler represents the first agent- aware service for multi-item negotiations 1 , since agent ser- vices have mostly focused on infrastructure issues related to negotiation protocols and ontologies. The paper is organised as follows. Section 2 introduces the market scenario where buyers and traders are to nego- tiate, along with the requirements, preferences, and con- straints they may need to express. Next, a formal model of the problem faced by the buyer (auctioneer) based on the de- scription in section 2 is provided. Thereafter, section 4 de- tails the computational realisation of the agent service as an agency. Finally, section 5 summarises our contributions. 2. Market requirements While in direct auctions, the items to be sold are physi- cally concrete (they do not allow configuration), in a nego- tiation involving highly customisable goods, buyers need to express relations and constraints between attributes of dif- ferent items. On the other hand, multiple sourcing is com- mon practice, either for safety reasons or because offer ag- gregation is needed to cope with high-volume demands. This introduces the need to express constraints on providers and on the contracts they may be awarded. Not forgetting 1 Awarded the Best Application prize of the 2003 Agentcities Technol- ogy Competition (http://www.agentcities.org). Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. AAMAS'04, July 19-23, 2004, New York, New York, USA. Copyright 2004 ACM 1-58113-864-4/04/0007...$5.00