Current challenges and trends in intelligent ... - Semantic Scholar

Knowl Inf Syst (2007) 12(2):117–127 DOI 10.1007/s10115-007-0069-3

Knowledge and Information Systems

EDITORIAL

Mat´ıas Alvarado · Leonid Sheremetov · ˜ René Banares-Alc´ antara · ´ Francisco Cantu-Ortiz

Current challenges and trends in intelligent computing and knowledge management in industry Received: 7 February 2006 / Revised: 20 November 2006 / Accepted: 25 November 2006 / Published online: 21 March 2007 C Springer-Verlag London Limited 2007

1 Introduction This special issue addresses a significant increase on research and applications of intelligent computing in industry in recent years. In psychology, intelligence is the general mental ability involved in calculating, reasoning, perceiving relationships and analogies, learning, storing and retrieving information, using language fluently, classifying, generalizing, and adjusting to new situations. According to Alfred Binet, intelligence is the totality of mental processes involved in adapting to the environment [6]. We understand by Intelligent Computing (IC), a combination of these intelligence facets involved in information processing and focused on providing intelligence to different aspects of computer applications. We use terms intelligent computing and artificial intelligence (AI) as synonyms of a discipline that generates tools that, in turn, provide intelligence, although they may be defined slightly differently in the technical literature. In most of the cases, we are also looking for the integration of these machine intelligence capabilities with human intelligence. In industry, these intelligent capabilities are always applied within some organizational context. Knowledge management (KM) embodies the organizational M. Alvarado (B) Departament of Computing, Centre of Research and Advanced Studies (CINVESTAV-IPN, Mexico), Mexico City, Mexico E-mail: [email protected] L. Sheremetov PIMAyC, Mexican Petroleum Institute, Mexico City, Mexico R. Bañares-Alcántara Department of Engineering Science, University of Oxford, Oxford, UK F. Cantú-Ortiz Tecnologico de Monterrey (ITESM), Monterrey, Mexico

118

M. Alvarado et al.

processes that seek a synergic combination of the data and information-processing capabilities of information technologies, with the creative and innovative abilities of human beings. In simpler terms, KM seeks to make the best use of knowledge that is available to an organization. In this sense, we consider KM solutions when both machine and human intelligent capabilities are applied within the organizational context. We stress that from our point of view both IC and KM are mostly integrative practices. This special issue brings together selected papers from leading research groups from different countries and continents (US, Mexico, UK, Spain, Russia, Singapore) discussing current trends of these technologies and based on their experience in developing industrial applications. These trends are considered from three different perspectives: current trends in automation from an industrial perspective; trends in IT methodologies to cope with industrial challenges; and a technological perspective, i.e. trying to find the best suitable technologies and technological frameworks supporting these methodologies. We try to follow these trends through cutting edge application examples from the automotive, chemical, oil & gas, and transportation industries covered in this special issue. 2 A new age of industrial automation In this section, we will try to answer the following question: what are the new demands on information technologies at the current stage of industrial automation? To answer, let us start with the main challenges in economy. Nowadays, industrial organizations have to constantly update market and organizational capacities to attain a competitive advantage, innovation, and agility in their business processes. Ann Macintosh [3] identified some of these specific business factors. We consider that associated with today’s market, these factors can be grouped within the two basic categories. The first group associated with the complexity and dynamicity of products and services is composed of the following factors: – Products and services are increasingly complex, endowing them with a significant information component. – There is a need to manage increasing complexity as small operating companies are transnational sourcing operations in supply chains. – This complexity implies collaborative participation of the diverse experts together with the feedback/trial-and-error cycle to adjust the solutions. – Companies look for more flexible organization structures to respond to the changes in the environment: networks operate at different levels. In a network, multiparty cooperative relationships and high degrees of flexibility are the keys. – Control of the processes must be flexible enough to account both for the dynamicity of the market, and for perturbations to the process and even major disruptions. – On the other hand, the second group is associated with the company’s knowhow and its role for the company’s competitiveness and success at the market. – Market places are increasingly competitive and the rate of innovation is rising. – Competitive pressures reduce the size of the work force that holds valuable business knowledge.

Current challenges and trends in intelligent computing

119

– Reductions in staffing create a need to replace informal knowledge with formal methods. – Changes in strategic direction may result in the loss of knowledge in a specific area. – The amount of time available to experience and acquire knowledge has diminished. – Early retirement and increasing mobility of the work force lead to loss of knowledge. We find many of these factors reflected in the papers of this special issue. Over the years, automation and the technological advancements have led to many industrial breakthroughs such as mass production, improved quality, reduced manufacturing lead times resulting in a more competitive industry and lower prices to the consumer. Modern industry shows evolving complexity of technical processes. For the refinery supply chain (SC) discussed in the paper “Artificial intelligence methodologies for agile refining: An overview” by R. Srinivasan, for example, this complexity is revealed starting with the available alternatives to obtain the raw materials, their transport and storage; the diversity of the chemical, mechanical, and physical process for refining, and the network of channels and conditions of oil, fuel, and gas distribution. As a consequence, traditional applications like CAD, process control or management have evolved due to considerable changes in design, operation, procurement, and management. The flexible process control and supply chain concepts have been applied successfully in the oil & gas industry. Management of the steps of supply, manufacturing or operation chain requires to automate the integration of the parts of the whole production, distribution logistics, and sales. As we see from the paper by R. Srinivasan, there are specialized systems in the oil industry to support from crude oil acquisition to terminal and depot management, thus cutting across the complex web of logistics, manufacturing, and sales. These systems have to enable the agility of the whole chain, the ability to proactively respond to a dynamic environment through responsiveness to customers, product and process innovation and operational flexibility. The flexible process control and supply chain concepts have also been applied successfully in Automotive industry as one of the leaders in this innovation process: in 2001, Ford Motor Company reported the largest R&D expenditures ($7.4 billion) among the private sector of the U.S. economy [5]. The paper “Intelligent systems in the automotive industry: Applications and trends” by O. Gusikhin, N. Rychtyckyj and D. Filev analyzes the evolution and the impact of the IC technologies in this industry. As we can see, the improvements started from the basic functions of in-vehicle intelligence, design and manufacturing processes and data mining in warranty analysis and evolved to the complex integrated solutions like vehicle assembly planning, assembly sequencing and production control as well as on introducing supply chain (SC) and corporate KM. In this century, telecommunications technology is resulting in organizational environments that are increasingly more global, complex, and connected. SC is a particular facet of such type of organizations. These new enterprise models have emerged as a response to market challenges. Virtual and networked organizations arise as alternatives providing services beyond the traditional business boundaries and scopes. In the paper “Ontology-driven intelligent service for

120

M. Alvarado et al.

configuration support in networked organizations” by A. Smirnov et al., the behaviour of a networked organization is compared to that of an intelligent organization. The latter “behaves as an open system which takes in information, material and energy from the environment, transforms these resources into knowledge, processes and structures that produce goods or services which are in turn consumed by the environment” [2]. Such behaviour assumes the existence of organization “intellectual abilities,” i.e. abilities to create, exchange, process, and infer knowledge. Knowledge and information are the medium in which business problems occur. Modeling of complex industrial and business processes addresses the classification and categorization of technical as well as business management information and the integration of all the information processes. That is why the business operations area is making a powerful push of IT with systems that integrate the different parts of the business. This new trend on industrial automation demands systems able to handle information and knowledge (I&K) in a more efficient and integral way. We call this trend an intelligent automation.

3 Intelligent and knowledge-based automation The question we would like to be answered in this section is as follows: what are the most suitable approaches to tackle the challenge of intelligent automation? First of all, we look for the methodologies and the deployment of tools resulting in agile technical processing. Methodologies drawn from traditional artificial intelligence—software agents, pattern recognition, expert systems—have a role to play in this path towards agility. As an example, we can consider design automation, which is increasingly applying knowledge-based engineering (KBE) concepts and methods as defined by Gousikhin et al., the use of computer tools to enhance the design process by bringing engineering design, manufacturing engineering, consumer-related and other appropriate knowledge up front in the creation process and, wherever possible, to automate the process. The application of KBE addresses the modeling of the reasoning process used by the design engineer and evaluates it against predetermined design criteria and standards. It also integrates symbolic and numeric computation. Knowledge-based decision support systems using techniques from AI and expert systems to provide smarter support for the decision-maker began evolving into the concept of organizational knowledge management about a decade ago, and are now beginning to mature. Increasingly overwhelming amount of available information, used for planning, decision-making, and optimization of industrial processes demands for this methodology in order to extract knowledge from huge and diverse volumes of information. KM offers, as well, agile methods to organize and reuse information for executing business procedures. Perhaps knowledge management is one of the main “hot” issues to be fully integrated as part of the cutting edge technologies providing added value to information technologies in industry. The use of knowledge management methodologies: – facilitates the whole design–deployment–application–procurement industrial life cycle,


121

– supports the successful transfer of knowledge from senior experts (engineers, designers, managers) to junior staff, – allows human experts to use knowledge skills and experience from others within the organization without sacrificing productivity and design quality. Additionally, preservation of expert know-how, together with the improvement of the effective usage of preserved knowledge, is strategic to promote sustainability of industrial services. In this special issue, KBE applications are illustrated, first through a framework for hierarchical redesign of chemical processes in the paper “A hierarchical approach for the redesign of chemical processes” by I. López-Arévalo et al., and then for pipe stress analysis in the paper “Decisionmaking on pipe stress analysis enabled by knowledge-based systems” by M. Alvarado et al. In both works, the reasoning framework works at different levels of expertise and generates alternative design solutions based on heuristic knowledge coupled with the input parameters provided by the design engineer. Design parameters are processed by specific domain software, usually a simulator. Design proposals are generated by combining general domain knowledge and knowledge specific to the problem at hand, which can be implemented in knowledge-, principle-, rule-, or simulation-based systems. The reasoning and simulation systems support human expert intervention when deciding upon the final design. Complementary to the preservation and reuse of know-how, the automation of a framework embodying decisions representation is also needed. This can be achieved by integrating information about the designed device, the workflow steps, and the mutual knowledge among participants in the process. The trace of the workflow steps keeps a record of the decision-making process. Furthermore, from a holistic perspective, the design of engineering systems on structure and actionable phases requires the usage of ontologies and languages of processes to manage the amount, complexity, and variety of information to be integrated. An information ontology can be described as a knowledge representation model that contains specific knowledge about a problem domain. The use of ontologies facilitates the importation of data descriptions from various data sources as well as mapping between the concepts and relations from these sources. The current challenge is to integrate and use knowledge bases and ontologies across the entire industry supply chain. This challenge embraces the modeling of the business rules, logic, and inference strategies that describe how the processes function in each problem domain. Moreover, knowledge-managed ontologies-based information should be used for planning, decision-making, and process optimization. Additionally to decision-making on pipe stress analysis and hierarchical design, the papers from this special issue illustrate KM application to supply chain management and operation management for nimble refineries. I&K intensive tasks require a special focus on information and knowledge logistics as illustrated in the paper “Knowledge and information distribution leveraged by intelligent agents” by R. Brena et al., and also in the paper by A. Smirnov et al. Knowledge logistics addresses a dynamic aspect of knowledge management: activities over the knowledge delivery. These activities concern acquisition, integration, and transfer of the right knowledge from distributed sources located in an information environment and its delivery in the right context, to the right person, in the right time for the right purpose.

122

M. Alvarado et al.

In conclusion, the deployment of intelligent computing methodologies represents a prime opportunity for achieving substantial savings, significant improvements in human performance, and competitive advantage. That is why in the first decade of the millennium, according to a study of the worldwide AI market performed by Business Communication Company (BCC), the latter is estimated in more than $21 billion by 2007 with an average annual growth rate from 2002 to 2007 of 12.2% [4]. 4 Technological frameworks Finally, this technology-centric section is meant to answer the following question: which are the main techniques proved or considered to be the most suitable to support intelligent automation? The industry of today is regularly adopting new technologies. The study we cited above targets five AI technologies as the fastest growing AI components: expert systems, belief networks, decision support systems, neural networks, and agents [4]. Future expectations are associated with Web intelligence based on the decentralized computing infrastructure available from autonomous software agents. Most of these technologies are referred to in the papers of this special issue. Technologies of IC are indispensable in an attempt to improve the management of complexity (technical, structural, and organizational) in industrial processes. There are several trends in information technology related to the management of complexity of technical systems: (i) the use of computational intelligence techniques, (ii) the use of agents to integrate methods of different computational nature (usually combining quantitative and qualitative models, soft and hard computing within the context of hybrid intelligent systems), (iii) human-centered technology searching for the synergy between human and computational intelligence, (iv) semantic integration of I&K, and finally, (v) the use of integrated frameworks and tools implemented over interoperable infrastructures. In this section, we analyze each trend briefly. Internal complexity of the set of attributes, parameters, and stages to be integrated in the modern technological processes requires the use of flexible techniques providing a holistic treatment of that complexity. A promising trend of research enabling the integration of the huge diversity of information to be processed as well as the openness to analyze scenarios and assess situations is the deployment of computational intelligence methodologies. The clear evidence of the proliferation of this trend can be found in the paper by O. Gusikhin et al., focusing on the applications of computational intelligence methodologies as building blocks of intelligent engineering systems providing intelligence to all aspects of automotive industry ranging from vehicle components up to supply chain management and corporate knowledge management. It is important to stress that computational intelligence methodologies (fuzzy logic, neural networks, evolutionary computation, machine learning, knowledge representation, probabilistic/possibilistic reasoning) have gained wide acceptance in industry due to the possibility to provide solutions for complex problems when the classical methods are infeasible, inefficient, or uneconomical. As we see, in many systems several techniques are used together or even go hand in hand with the traditional methods. In recent years, hybrid intelligent


123

systems, integrating different AI techniques from neural networks to fuzzy logic, have made solid steps towards becoming more accepted in mainstream industry due to their capabilities to handle real-world complexities such as imprecision, uncertainty, and vagueness [8]. For example, in the paper by R. Srinivasan discussing operations and SC management, hybrid integrated models are composed of different techniques: intelligent models using an agent-based approach, expert systems, dynamic time warping, data mining (trend analysis), which contrast with model-based techniques using neural networks and dynamic principal components analysis. In the paper by O. Gusikhin et al., expert systems, data mining, language processing, and agents are used for knowledge-based engineering and process planning; while agents and ontologies are used for SC and corporate KM. Agent technology is seen as an agglutinating framework for models of different nature. The benefits of the agent technology in contrast to the commonly used mathematical programming is that it is more versatile and can easily capture the qualitative as well as the quantitative nature of a complex technical system. The autonomous agent applications in industry carry on the usage of AI methods to deal with diverse information in a flexible manner. Examples of the application of agent technology can be found in nearly all the papers of this issue. They explore different AI techniques like multi-valued logics, genetic algorithms, and neural networks for codification of autonomous agents’ knowledge as well as the reasoning capabilities. Agents are by no means monolithic pieces of knowledge, but rather encompass many aspects such as communication, coordination techniques, cooperation, autonomy, competition, distribution, and mobility, among others. These particular capabilities offer the promise of increased modularity, flexibility, and adaptability when building complex systems, particularly multiagent systems. Another central point for all the approaches described in this special issue is that human intervention is usually required, i.e. most of the intelligent systems discussed in this issue aim to support and not substitute humans, trying to achieve synergy between human and computational intelligence. To improve the management of complexity during the redesign of technical processes in the paper by I. Lopez et al., a model-based reasoning is proposed to automatically generate alternative representations of an existing process at several levels of abstraction; this step results in sets of units or meta-units organized according to their functions and intentions. A case-based reasoning system then retrieves from a library of cases units or meta-units similar to the one selected by the user, according to the function and intention of a unit and its neighbors. The final output is a set of cases (units or meta-units) ordered according to their similarity. Human designer provides the input to the system and interprets the results, adapting the most promising case within the original process. An actor playing a role at the beginning of design and redesign is the simulator that is a software tool used to obtain the design description of the process together with the functional analysis that is modeled using the data extracted, the simulator is also used to implement and evaluate the generated alternative process designs. Then the reasoner, case-based in the paper by López et al. or rule-based in the paper by Alvarado et al., identifies the suitable equipment/section to be modified and generates more adequate equipment/sections based on the new requirements; as the next stage, using the redesign criteria and human design expertise, the

124

M. Alvarado et al.

reasoner generates alternatives based on the identified candidates, and evaluates them such that the most similar/promising one is ready to be adapted by the user into the original process based on the evaluation of its performance within the overall process. This is an iterative cycle that finishes when an appropriate alternative process design is obtained according to the new objectives. For an industrial enterprise, it is of decisive strategic importance to optimize the information flows and to implement an efficient reuse of existing knowledge. In knowledge-intensive industries, information is a major production factor and knowledge reflects an important asset of the enterprise. Several papers from this issue tackle the problem of integration and sharing of I&K available from heterogeneous data sources, such as databases, Web sites, and specialized data servers. They show that the functionality and usability of information systems can be augmented by incorporating knowledge representation and inference techniques from AI. From the papers by A. Smirnov et al., I. López-Arévalo et al., and R. Brena et al., we can see a new dimension of information and knowledge integration reached with the use of ontologies for semantic information management. In order to achieve semantic interoperability in heterogeneous information systems (such as concurrent engineering or corporate knowledge management), the meaning of the information that is interchanged has to be understood across the systems. The use of ontologies for the explication of implicit and hidden knowledge is a possible approach to overcome the problem of semantic heterogeneity [7]. Ontological modeling is one of the technologies playing a key role both in Web-empowered and knowledge-intensive applications widely covered in this issue. Ontologies set up languages that organize information around the relevant aspects of the industrial domain like the information on workflow, tasks, and goals of the processes. The dynamic information is being modeled by languages of business processes capturing the complex relationships among the parts of the process. In the paper by O. Gusikhin et al., the successful application of ontologies in the automotive industry is illustrated through the ability to read and understand free-form text. The method for dealing with unstructured text developed at Ford Motor Company contains a mechanism for the automated creation of rules that are then utilized to deliver targeted unstructured information. This system reduces the amount of time that must be spent by manufacturing engineers that are responsible for maintaining manufacturing engineering knowledge. Industrial applications are no longer solid monolithic constructs; they usually contain numerous heterogeneous components involved in complex interactions within dynamic organizational contexts. The breakthrough technologies to handle this type of complexity are Web intelligence and agents. Web intelligence deals with the advancement of Web-empowered systems, services, and environments. It includes issues of Web-based knowledge processing and management; distributed inference; information exchange and knowledge sharing [9]. The papers of this special issue by A. Smirnov et al. and R. Brena et al. propose technological ontology-driven frameworks based on very similar principles of Web intelligence. Ontological models form the main constituent element of these technological frameworks (KSNet, JITIK, PROSIS) providing a common knowledge representation for integrated frameworks and tools. These tools use Web services widely, an emerging technology enabling application integration in the enterprise information system in a service-oriented fashion. These architectures support loosely


125

coupled services to enable business flexibility in an interoperable technology that allows the integration of components without the need to know which technology is behind each of them, adapting to the component’s technology without changing its original functionality [1]. 5 Concluding remarks Papers selected for this special issue cover different aspects of intelligent computing and KM for industrial intelligent automation. We believe that systems, incorporating the synergy of human and computational intelligence along with the hybrid schemes within their functional components, will be able to make significant steps towards changing the “black box” image that has been associated with several AI-related techniques and bring it closer to a “transparent box” enhancing its acceptance by practitioners. The use of AI and intelligent systems in industry is more prevalent than we could discuss in this special issue. This set of articles does not necessarily illuminate everything that is happening in intelligent automation, but aims to take a step towards our understanding of the real potential of intelligent computing for industrial automation. We look forward to continued work in this area and to the day when the intelligent computing serves us more actively and with more enlightenment. Acknowledgements Finally, we would like to thank all the contributing authors for excellent research papers, their institutions for the support given for research work in the field of intelligent computing, as well as the hard-working reviewers who gave their valuable time to ensure the high quality of the KAIS papers. Special thanks to the KAIS Editorial Board for their interest in the topic and for providing the possibility to publish this special issue.

References 1. Borges B, Holley K, Arsanjani A (2004) Service-oriented architecture. http://webservices. sys-con.com/read/46175.htm. Cited Aug 2004 2. Choo ChW (1998) Information management for the intelligent organization, 2nd edn. Information Today, Medford, NJ 3. Macintosh A (1997) Position paper on knowledge asset management. Artificial Intelligence Applications Institute (AIAI), Edinburgh 4. McKellar H (2003) Artificial intelligence: past and future. Creating and managing the knowledge-based enterprise. KMWorld 12(4) 5. NSF (2004) Science and engineering indicators, http://www.nsf.gov/statistics/seind04/c4 /c4s1.htm 6. The Columbia Encyclopedia, 6th edn (2001) 7. Wache H, Vogele T, Visser U, Stuckenschmidt H, Schuster G, Neumann H, Hubner S (2001) Ontology-based integration of information – a survey of existing approaches. In: Stuckenschmidt H (ed) IJCAI-01 workshop: ontologies and information sharing, pp 108–117 8. Zhang Z, Zhang Ch (eds) (2004) Agent-based hybrid intelligent systems: an agent-based framework for complex problem solving. LNAI 2938(XV):196 p 9. Zhong N, Liu J, Yao Y (2002) In search of the wisdom web. Computer 35(11):27–31

126

M. Alvarado et al.

Author Biographies

Mat´ıas Alvarado is currently a Research Scientist at the Centre of Research and Advanced Studies (CINVESTAVIPN, Mexico). He got a Ph.D. degree in computer science from the Technical University of Catalonia, with a major in artificial intelligence. He received the B.Sc. degree in mathematics from the National Autonomous University of Mexico. His interests in research and technological applications include knowledge management and decision making; autonomous agents and multiagent systems for supply chain disruption management; concurrency control, pattern recognition and computational logic. He is the author of about 50 scientific papers, a Journal Special Issues Guest Editor on topics of artificial intelligence and knowledge management for the oil industry; an academic, invited to the National University of Singapore, Technical University of Catalonia, University of Oxford, University of Utrecht, and Benemérita Universidad Autónoma de Puebla.

Leonid Sheremetov received the Ph.D. degree in computer science in 1990 from St. Petersburg Institute for Informatics and Automation of the Russian Academy of Sciences, where he has worked as a Research Fellow and a Senior Research Fellow from 1982. Now he is a Principal Investigator of the Research Program on Applied Mathematics and Computing of the Mexican Petroleum Institute, where he leads the Distributed Intelligent Systems Group, and a part-time professor of the Artificial Intelligence Laboratory of the Centre for Computing Research of the National Polytechnic Institute (CIC-IPN), Mexico. His current research interests include multiagent systems, semantic WEB, decision support systems, and enterprise information integration. His group developed CAPNET agent platform and has been involved in several projects for the energy industry ranging from petroleum exploration and production to knowledge management with special focus on industrial exploitation of agent technology. He is also a member of the Editorial Boards of several journals.


127

˜ René Banares-Alc´ antara has worked in the University of Oxford from October 2003 and is now a Reader in engineering science at the Department of Engineering Science and a Fellow in engineering at New College. He previously held a readership at the University of Edinburgh and lectureships in Spain and at the Universidad Nacional Autónoma de México (UNAM). He obtained his undergraduate degree from UNAM and the M.S. and Ph.D. degrees from Carnegie Mellon University (CMU). Starting with his work at CMU, his research interests have been in the area of process systems engineering, in particular chemical process design and synthesis. He has developed a strong relationship with computer science/artificial intelligence research groups in different universities and research institutes, with current research also linking to social and biological modeling. He has (co)authored more than 100 refereed publications and has been a Principal Investigator and a Researcher in several EPSRC and European Union projects.

´ Francisco Cantu-Ortiz obtained the Ph.D. degree in artificial intelligence from the University of Edinburgh, United Kingdom and the Bachelor’s degree in computer systems engineering from the Tecnológico de Monterrey (ITESM), México. He is a Full Professor of artificial intelligence at Tecnológico de Monterey and is also the Dean of research and graduate Studies. He has been the Head of the Center for Artificial Intelligence and of the Informatics Research Center. Dr. Cantú-Ortiz has been the General Chair of about 15 international conferences in artificial intelligence and expert system and was a Local Chair of the International Joint Conference on Artificial Intelligence in 2003. His research interests include knowledge based systems and inference, machine learning, and data mining using Bayesian and statistical techniques for business intelligence, technology management, and entrepreneurial science. More recently, his interests have extended to epistemology and philosophy of science. He was the President of the Mexican Society for Artificial Intelligence and is a member of the IEEE Computer Society and the ACM.