1: Keynotes

• Understanding the Service Revolution, Prof. Roland T. Rust, University of Maryland
  Abstract: Service is creating a paradigm shift in marketing. This presentation provides the underlying causes of this paradigm shift and identifies the central role of technology. The implications of further technological development are explored and the consequences for the future of marketing are investigated.
  
  
• Services: The Other 75% of the Economy! Prof. Richard Larson, Immediate Past President of INFORMS, MIT
  Abstract: As nations grow and develop more as knowledge-based economies, services industries become a higher and higher fraction of the GDP. In the USA and some European economies, services comprise 75% of the GDP. While those figures do not yet represent most Asian economies, some of the fastest growing businesses in China, India and elsewhere in Asia are services businesses. In this presentation we focus on how a broadly-based engineering analysis of services, buttressed by principals of social sciences and management science, can lead to vastly improved designs and operations of services industries. We provide as illustrations various services in health care, technology-enabled education (including e-learning) and catastrophe response (including natural disasters such as hurricanes and typhoons, earthquakes and influenza pandemics).
  
  
• Using Mathematics to Solve Service Operations, Industrial, and Logistics Problems, Prof. Kathryn E. Stecke, University of Texas at Dallas, USA
  Abstract: Mathematics has been called the language of science. Mathematics is used to solve many real-world problems in service operations, industry, logistics, the physical sciences, economics, social and human sciences, engineering, and technology, for example. We overview the many service, industrial, and logistics problems that have been solved using fuzzy logic technology, multiobjective metaheuristics, neural networks, tabu search, genetic algorithms, simulated annealing, mathematical programming, decision analysis, Petri nets, and queueing models. This overview could be useful for new faculty and Ph.D. students in Industrial Engineering, Operations Research, and Operations Management who are looking to solve some real problems. Future applications are also described.
  

2. Workshop Sessions

• “Stochastic Decomposition for Transshipment Modeling” By Suvrajeet Sen, University of Arizona
  Abstract: Transshipment problems deal with situations in which demand at distributed locations can be met by balancing the distribution of supply among network nodes where there is a mismatch betweeen supply and demand. Such mismatch usually results from demand uncertainty at each node of the network. This problem has been traditionally solved using simulation- optimization methods. We will show that stochastic decomposition provides a very powerful apporach for such problems because of the ability to use the network structure of the problem, and to use stopping rules that yield reliable solutions. (Co-author: Lei Zhao, Tsinghua University)
  
  
• “Production Planning Under Supply and Demand Uncertainty: A Stochastic Programming Approach” by Julia Higle, Ohio State University
  Abstract: Motivated by problems that arise in semi-conductor manufacturing, we consider the problem of producing lots to satisfy demand under conditions leading to supply uncertainty (e.g., throughput times and yields) as well as demand uncertainty (e.g., quantities and due dates). Using data inspired by actual operating conditions, we will discuss preliminary computational results and insights obtained. (Tom Prevendoski, Ohio State University, Karl Kempf, Intel Corporation)
  
  
• “Data Mining, Forecasting, and Activity Monitoring” by Kwok-Leung Tsui, Georgia Institute of Technology
  Abstract: This talk discusses strategies and techniques in data mining, forecasting, and activity monitoring. Data Mining refers to non-trivial extraction of knowledge from large volume of data. Activity monitoring refers to detection of interesting events that require actions (e.g., detection of customer churn, credit card or insurance fraud, and computer intrusion). Forecasting refers to predicting future activities based on historical data and other variables (e.g., demand forecasting, sales forecasting, stock price forecasting). We will propose a general strategy for modeling, monitoring, and forecasting of dynamic systems. In particular, we will discuss a statistical process control approach for business activity monitoring. We will also proposal a churn detection procedure for customer profile modeling. Several examples and case studies in telecom and service industries will be used to illustrate the proposed methods. (Co-author, Wei Jiang, Steven Institute of Technology)
  
  
• “Using Nested Partitions for Beam Angle Selection in Intensity-Modulated Radiation Therapy” by Leyuan Shi, University of Wisconsin – Madison
  Abstract: Modern treatment technologies allow clinicians to develop complex treatment plans for a wide array of illnesses, including many forms of cancer. While expert judgment may lead to good treatment plans in many cases, computer automated decision support tools provide a mechanism by which enormous numbers of alternative plans can be automatically generated and compared, and thus yield treatments that are often significantly better. In this talk, we describe results obtained by applying the Nested Partitions (NP) metaheuristic to the problem of selecting beam angles for radiation delivery in Intensity Modulated Radiation Therapy.
  
  
• “Information Technology as a Service” by Robin Qiu, Pennsylvania State University
  Abstract: In this information era both business and living communities are truly information technology (IT) driven and service-oriented. As the globalization of the world economy gets accelerated with the fast advance of networking and computing technologies, IT plays more and more critical role in assuring the real time collaborations of delivering needs across the world. Nowadays the world-class enterprises are eagerly embracing the service-led business models aimed at creating highly profitable service-oriented businesses. They take advantage of their own unique and years marketing, engineering, and application expertise and shift gears towards creating superior outcomes to best meet their customers’ needs in order to stay competitive. IT has been considered as one of the high value services areas. In this talk, the discussion will focus on IT as a service. We present IT development, research, and outsourcing as a knowledge service; on the other hand, we argue IT as a service helping enterprises align their business operations, workforce, and technologies to maximizing their profits by continuously improving their performances. Numerous research and development aspects of service enterprise engineering from a business perspective will be briefly explored, and then computing methodologies and technologies to enable adaptive enterprise service computing in support of service enterprise engineering will be simply studied and analyzed. Finally, future development and research avenues in this emerging interdisciplinary field will also be highlighted.
  
  
• “Role of Manufacturing as a service operation and Service Engineering Program at Penn State” by Sanjay Joshi , Pennsylvania State University
  
  
• “Resource and Demand Allocation in Multi-Site Service Systems with Inter-Site Customer Flows” by Xiuli Chao, North Carolina State University
  Abstract: Healthcare management is becoming a major problem in both developed and developing countries. As living standard improves, public is increasingly demanding high customer service levels. This talk reports on a study of strategic optimal resource allocation in healthcare environment. We formulate the problem as a multi-site service systems with inter-site customer flows, and develop analytical models and from which we obtain explicit closed form optimal allocation policy. We aim at providing insights to and guidelines for resource allocation in these service systems when some service criterion, such as average waiting time, loss rate, or blocking probability, is a major concern. Our results demonstrate that the commonly used proportional allocation rules are not optimal; instead the optimal resource allocation solution exhibits a structure of "a few large and many small''. The analysis involves queueing theory, game theory, convex analysis, and mathematical program with equilibrium constraints (MPEC). (This is a joint work with L. Liu and S. Zheng)
  
  
• “Port Supply Chains as Social Networks” by Peggy Lee, Pennsylvania State University
  Abstract: This paper demonstrates the use of social network theory in understanding the degree of coordination in port supply chains. Three port supply chains are used to illustrate that port supply chains are, in fact, social networks and that their success is strongly influenced by the level of coordination and cohesion in their networks. Of the three port supply chains in this study, those with higher degrees of cohesion and coordination tend to exhibit higher operational performance. While causation was not demonstrated, this paper provides the foundation and groundwork for further empirical study into the use of social network theory to study supply chain relationships and port supply chain integration.
  
  
• "Fault Injection-based Test Case Generation for SOA-oriented Software" by Jia Zhang, Northern Illinois University
  Abstract: The concept of Service Oriented Architecture (SOA) implies a rapid construction of a software system with components as published Web services. How to effectively and efficiently test and assess available Web services with similar functionalities published by different service providers remains a challenge. In this paper, we present a step-by-step fault injection-based automatic test case generation approach. Preliminary test results are also reported.
  
  
• “RFID – The Wireless Internet of Artifacts” by Rajit Gadh, UCLA
  Abstract: The last twenty-five years have marked the coming of the personal computing and communication industry. Result: Individuals now carrying devices that are personal, mobile and always connected to the Internet. It is my belief that in the next twenty-five years, such information carrying and disseminating capability will extend from the "computing/communication devices" to real-world non-computing artifacts that we use in every day life such as clothes, utensils, furniture, packages, etc. These artifacts will collectively form what I refer to as the "Internet of Artifacts", an idea whose time has come. Like with any grand challenge, these artifacts will need to be uniquely identified (using technologies such as RFID), they will need to communicate with each other (wirelessly) and will gradually have the ability to take intelligent decisions first individually and then collectively.

  At UCLA's WINMEC (Wireless Internet for the Mobile Enterprise Consortium), the Wireless "Internet of Artifacts" notion is being explored via a project called WinRFID (http://winmec.ucla.edu/rfid) -- which is the first generation of our implementation of this idea. RFID or Radio Frequency Identification is a technology that can embody the identity and other related information of an artifact within a chip called a tag that has no power source and make such information available to an RFID transceiver when the tag receives the RF transmission and its coupled energy. RFID tags are expected to eventually be embedded into every daily-life artifact. At UCLA, we are developing the WinRFID Middleware that allows efficient, intelligent and optimized networking and management of RFID readers, tags and sensors at the edge of the network. The WinRFID middleware is currently being used for several research and industrial-led projects at UCLA-WINMEC that include securing assets, asset tracking, managing object shipments in supply chains, factory wireless networks, etc.

• “Transforming Healthcare: Challenges and Opportunities” by Francois Sainfort , Georgia Institute of Technology
  Abstract: Dr. Sainfort will present challenges and opportunities related to applying systems engineering and management sciences approaches to transforming health care delivery systems from a current ineffective, reactive, disease-focused system to a future cost-effective, proactive, health- and wellness-focused system. Dr. Sainfort will provide background on the healthcare industry, present possible solution approaches and describe challenges and opportunities.
  
  
• "Trilinear Methods and Multiblock PCA for Improved Data Understanding" by Michael Piovoso, Pennsylvania State University
  Abstract: Trilinear data occur where data are collected for a system over a fixed time interval. Examples include parts manufacturing and chemical batch processes. These data are three dimensional in nature; there are variables over time over parts or batches. Traditional techniques for analysis include multiway PCA whereby the data are unfolded into a two-dimensional structure and then a standard PCA is applied. Trilinear methods involve a PCA like decomposition of the three-dimensional data. Two different approaches are most common: PARAFAC (Parallel factors) and Tucker-3 decomposition. PARAFAC finds n factors in each of the dimensions and approximates the three dimensional matrix into the sum of the outer products of these factors. Tucker-3 is a decomposition in which different numbers of factors are possible for each dimension. This decomposition also includes a smaller three-dimensional weighting matrix.

  Hierarchical and Consensus PCA are multiblock techniques that are useful when data come from different subunits and the effects of the subunits are of interest. For example, if in the manufacturing of a part there are various stages that must be executed before the product is completed, one may wish to understand the effects of each subunit on the product. Multiblock PCA may prove helpful. Hierarchical and Consensus produce a PCA-like models of the data for each subunit and a supermodel for the combined effects.

  This talk with introduce the trilinear and multiblock concepts and illustrate their application with an example from data on a batch chemical process. The purpose of this study was to determine the utility of trilinear and hierarchical methods for automatic outlier identification of process data.

• "An On-line Probabilistic Paradigm for Optimal Disassembly Planning" by Gina Tang , Rowan University
  Abstract: Disassembly is of growing importance in material and product recovery. However, the deployment of this process is complicated due to the lack of a priori information necessary for its control and planning. This paper develops a predictive model to tackle this problem..

3. US & Chinese NSF, IBM, and Sun Services Science Forum

A special forum was held from 3:30 PM to 5:10 PM on Juen 23, 2006. Attendees from US & Chinese NSF, IBM, and Sun delegations had open discussions on services science education and research issues.