Embedding Fuzzy Criteria in Request-for-Quote Automated Procurement Processes Miguel-Ángel Sicilia, David Palomar, Ignacio Aedo, Paloma Díaz DEI Laboratory, Computer Science Department, Carlos III University. Av. Universidad, 30 28911 Leganés (Madrid), Spain {msicilia, dpalomar, pdp}@inf.uc3m.es, aedo@ia.uc3m.es Elena García Computer Science Department, University of Alcalá. Ctra. Madrid-Barcelona, Km. 33,600 28871 Alcalá de Henares (Madrid), Spain elena.garciab@uah.es Abstract. The information-level integration of suppliers in manufacturing processes becomes a critical requirement in mass customization cycles, where the increasing wealth of product variants may force the production of units of size one. As a consequence, transactional procurement activities should be carried out with the help of flexible automated support tools, in order to adjust production costs, qualities and delivery rates. Request-for-quote (RFQ) scenarios in transactional procurement activities can be automated with current standards-enabled asynchronous business integration software frameworks, by using general-purpose message-passing mechanisms. Nonetheless, this automation leads to winning-quote selection criteria that could be considered too rigid in some cases, especially when using compensatory decision-making models. In this paper, we describe how simple fuzzy criteria – described through linguistic labels – can be integrated in the RFQ process to enable more flexible requisition specification and selection rules, and how existing business integration XML specifications can be extended to include that fuzzy information. 1. Introduction The practical implementation of mass customization relies – among other factors – on the potential offered by new technologies in information management, as described in (Piller et al, 2000). The coordination with the external suppliers involved in the individual prefabrication is considered one of the information costs that can be reduced through the use of these technologies. Specifically, Internet technology-based business integration can be considered an important enabler in order to achieve cost-effective hard customization methods (Piller, 2002). Moreover, mass customization leads to small lot sizes and may enlarge the complexity of logistics management (Knolmayer, 2001), and it also may increase the volume of their associated procurement interactions. In consequence, flexible fully automated or semi-automated procurement processes are a key technological infrastructure in efficient mass customization information cycles. Procurement tasks can be categorized as long-term- oriented (strategic) or short term-oriented (transactional). Strategic procurement activities include sourcing processes, identifying vendors and establishing and managing supplier relationships (Gebauer et al 1998). After a stable infrastructure has been established at the strategic level, a large number of individual buying operations occur at the transactional level. In this work, we focus on transactional request-for-quote (RFQ) procurement activities and their automation with software integration tools. Although these activities are sometimes considered as direct strategic ones (Gebauer & Segev 2001), they can also take place at the transactional level. At this level, in many cases the RFQ process could – and ideally should – be almost completely automated by establishing a number of quote selection criteria, which can optionally be carried on the business document itself. Quote selection criteria in natural language are seldom strict, since in many cases it’s difficult to express them through crisp sentences regarding price or other quote attributes. For example, sometimes only a vague price criterion is in the mind of the RFQ issuer, like “around a thousand dollars” or “not very expensive”. To allow for expressing those vague criteria in a direct way, a flexible criteria specification language is needed beyond the numeric and character types available in common business document schemas like XML SCHEMAS (W3C 2001). In this paper, we describe an approach to how this vagueness in quote criteria can be integrated with the business document itself, enabling their automatic communication through existing business-to-business (B2B) software integration engines. Our approach uses linguistic expressions based on fuzzy logic concepts to specify those criteria. Fuzzy logic has been successfully