INTERNATIONAL CONFERENCE ON ENGINEERING DESIGN, ICED11 15 - 18 AUGUST 2011, TECHNICAL UNIVERSITY OF DENMARK ICED11 1 KNOWLEDGE REPRESENTATION FOR SUPPLIER DISCOVERY IN DISTRIBUTED MANUFACTURING Christian McArthus, Farhad Ameri Department of Engineering Technology Texas State University San Marcos, TX 78666 ABSTRACT Online outsourcing has recently gained popularity among small and medium sized manufacturing companies as an efficient method for building flexible network of manufacturing counterparts. Several electronic marketplaces have emerged within the last few years with the objective of enabling large communities of buyers and sellers to virtually meet and establish new partnerships. Although e- marketplaces typically provide different automated search capabilities, they mainly rely on human users for final screening and evaluation of qualified suppliers. As the size of supply and demand pools increase, human-based search becomes inefficient. This paper describes an effort for enhancing the automation capabilities of web-based markets through an ontological approach. The proposed ontology is referred to as Manufacturing Service Description Language. MSDL provides formal semantic for manufacturing knowledge representation, thus enabling machine agents to actively participate in supplier discovery process. Keywords: Manufacturing outsourcing, ontology, semantic search 1. INTRODUCTION Web-based marketplaces for manufacturing services are currently the state-of-the-art in developing flexible supply networks for discrete part manufacturing [1, 2]. The growing popularity of online markets for manufacturing services can be attributed to several factors such as low cost of entrance, low cost of transaction due to elimination of market mediators, the possibility of interacting with a far larger number of potential counterparts, and equal treatment of members regardless of their size and global reach[3]. ChemConnect 1 and Covisint 2 are examples of online markets, in the chemical and automotive industries respectively, which evolved rapidly within a few years and now provide e- commerce services for several thousands of companies in more than 150 countries. The general architecture of online markets for manufacturing services is depicted in Figure 1. Using the bidding system, sellers bid on selected RFQs (request for quote) available in the RFQ database. Buyers evaluate the bids and select the ones which best meet their requirements. Online RFQ markets usually provide traditional means of search including keyword search, directory search, and database search on both buyer and seller data. Search criteria for querying the buyers database typically include customer’s category (process, sub-process, and product), materials, delivery location, industry, quantities (min-max), and dimensions. On the seller’s side, search criteria often incl ude process, sub- process, company name and location, and quality certification. These criteria, however, are simplistic and often provide incomplete picture of supplier’s potential match with requirements, leading to identification of suppliers that are irrelevant. Therefore, to arrive at more accurate results, the output of the search engine is further refined by human users through reviewing the narrative description of suppliers’ capabilities provided in a free-text format. However, as the size of the search space increases, human-based evaluation and screening becomes increasingly inefficient and error-prone. 1 http://www.chemconnect.com/ 2 http://www.covisint.com/