Fritz, M., Schiefer, G. (2002). Market Monitoring in Dynamic Supply Networks and Chains: an Internet-Based Support System for the Agri-Food Sector. Journal on Chain and Network Science 2 (2): 93-100
Market Monitoring in Dynamic Supply Networks and Chains: an Internet-Based Support System for the Agri-Food Sector Melanie Fritz, Gerhard Schiefer University of Bonn, Germany
Authors’ Address Dept. of Agricultural Economics, University of Bonn Meckenheimer Allee 174, D-53115 Bonn, Germany Phone: +49-228-733507. Fax: +49-228-733431 e-mail:
[email protected];
[email protected]
Market Monitoring in Dynamic Supply Networks and Chains: an Internet-Based Support System for the Agri-Food Sector
Abstract: Enterprises in dynamic supply networks and chains with changing business partnerships are involved in complex dependency structures and interrelationships. Consequently, awareness of developments in the competitive and market environment is of paramount importance for economic success. The research presented in this paper deals with the conception of Internet-based support systems which supply actors in dynamic supply networks and chains with individualized information about their competitive and market environment. The paper builds on the situation in the agri-food sector but the discussions of the theoretical background and the principal system design generally apply to dynamic network scenarios. The paper presents a conceptual approach for a market monitoring system, which integrates results of research from a variety of scientific fields, utilizes focused rule-based expert knowledge, and implements research supported automation potentials. The design allows for information quality, efficiency and economic feasibility. First empirical results from experimental studies support the concept. The paper concludes with the discussion of a research outline for further validation initiatives in experimental and sector environments.
Key words: market monitoring, food supply chains, networks, environmental scanning 1. Introduction
Enterprises in any sector of the economy are essential parts of interlinked production, trading and service processes which are commonly referred to as supply chains (e.g., Copacino, 1997). The agri-food sector is no exception. However, the sector’s infrastructure with its majority of small and medium sized enterprises, especially in early stages of the food production processes, together with particulars of food production generates complex dynamic dependency structures. Enterprises are embedded in
2 dynamic networks of relationships, in which they might act as integral parts of temporarily evolving supply chains with high variability in participation. In such situations, constant awareness of changes in the dynamic competitive and market environment becomes a crucial element for economic success. A prominent source for information about competition and markets is the vast number of distributed data sources provided through the Internet. However, a satisfactory and efficient use of these sources requires a focused and systematic scanning of their content and the linkage of scanning procedures and scanning results with the specific needs of enterprises in dynamic network infrastructures. In this paper, we suggest a dynamic, Internet-based and individualised sector monitoring management information system (DYS-MIS) to assist network and chain actors in their scanning activities in the dynamic information environment of the Internet and to provide them with a continuous flow of information on their competitive and market environment. The focus of the paper is on the logical deduction of an appropriate system framework, the transfer of the framework into operational system components, and the integration of the components into a comprehensive scanning and monitoring system. The paper is organized as follows: it first introduces the application environment of dynamic supply chain and network relationships, which constitutes the particular need for the continuous monitoring of competition, markets and market policies (paragraph 2). It then outlines the principal organizational components of an appropriate scanning and monitoring process (paragraph 3), relates them to relevant domains of scientific research, and formulates a conceptual framework for the dynamic monitoring system. The transfer of the framework into a sector scanning system (paragraph 4) requires the deduction and sequencing of operational activities. Expert knowledge, its integration through well-defined process routines, its utilization for process automation and its subsequent empirical evaluation and improvement are key features in a variety of activities. The system’s validation, performance, and adaptation to sector specifics is the focus of the final paragraphs. They provide an overview on first empirical results and outline proposals for further research on the design and organization of systems which meet the
3 market information needs of enterprises in designated dynamic network environments of the agri-food sector. 2. Application environment
Market situations with a dynamic network of enterprise relationships and temporarily evolving and altering supply chain configurations differ from traditional supply chain views (e.g., Tan, 2002) by the magnitude of potential linkages between actors and the variability of supply chain linkages that might evolve anytime depending on actual developments in markets, market policies or competition activities. While such dynamic supply networks are more the rule than the exception in the agrifood sector (e.g., Ménard, 1996), their organizational problems have received much less attention in research than the organizational problems faced by supply chains with a clearly defined and stable membership (for an example see, e.g., van der Vorst, 2000). In dynamic networks, potential needs of actors to change existing relationships, the necessity to respond to changes in network relationships and market activities of others and the need for trust building activities within a variable network structure require, in principle, from each actor an understanding of every potential supplier or customer in the network at every level of the production chain, even if no current relationship exists. This increase in information complexity is mirrored by information requirements on the technological, socio-economical and political environment. The situation is outlined in figure 1. The network linkages of enterprises within and between different network levels and the embedding of the network in its environment depict the complexity of enterprises’ principal requirements on infrastructures for communication and information delivery. It is apparent that enterprises within a common sector environment have, to some extent, matching market related information interests. In this situation, feasibility and efficiency concerns ask for a system design where the scanning of the competitive environment builds on joint activities of some or all of the enterprises in the network, whereas the needs of individual enterprises ask for individualized and flexible information delivery, which easily adapts to changing network situations.
4
Actual Supply Chain Within Dynamic Enterprise Network FARMS
Technology situation
INFORMATION ENVIRONMENT
Social situation
PROCESSORS
Network Levels
Political situation
TRADERS Economic situation
CONSUMERS
Figure 1: Information environment of enterprises in dynamic network situations.
Both requirements can be combined in dynamic sector monitoring systems where individualization and flexibility in delivery for enterprise needs are provided through appropriate features for personalization, allocation, and filtering of information. With the distributed information bases of the Internet as the primary source for information scanning activities, the approach combines sector and enterprise views in a dynamic, Internet-based, and individualised sector monitoring management information system (DYS-MIS). 3. Conceptual framework
Literature does not present a comprehensive framework for the development of dynamic systems of the type described above. This lack of a focused theoretical basis requires the development of a conceptual framework that links relevant sets of distributed scientific and expert knowledge to principal design requirements of a dynamic DYS-MIS system.
5 The conceptual framework presented in this paper defines the actors’ external information needs regarding developments in markets, market policies and competition activities as the central problem to be solved and accentuates it as starting point for further considerations regarding the - identification of system components that contribute to its solution, - determination of relationships among these components, and - identification of scientific knowledge bases for their organization. The system components and their relationships establish an information circle (figure 2), which describes the logical sequence of necessary steps towards the satisfaction of chain and network actors’ information needs. The principle outline of the circle resembles classical management information processes (see, e.g., Wigand et al., 1997). However, for the chain and network environment and with focus on dynamic monitoring of market related developments, the organization of the system components and their relationships is more complex and depends on the availability of appropriate scientific knowledge bases.
Information needs Information provision
Inform. personalization Information mapping
Inform. needs analysis Inform. sources identification Information collection
Figure 2: The DYS-MIS information circle with principal system components.
The conceptual framework links the different components to existing domains of scientific research (table 1). The diversity of backgrounds has limited interaction between the identified domains of research and excluded a coordinated focus towards dynamic monitoring systems for network environments. The analysis of external information needs of enterprise management has been the explicit focus of the ‘critical success factors’ method. It deduces critical and necessary
6 information from the areas in which ‘things must go right’ for an enterprise to be successful. To avoid deficiencies in the identification of information needs, the management-driven critical success factors analysis should be combined with a thorough analysis of market competition as demonstrated by Porter’s ‘five forces for industry analysis’. The determination of appropriate information sources needs to draw on a variety of research domains, which complement each other. ‘Business information’ deals with the relevance of information content and possible sources, ‘information search’ with the process of information seeking and its influencing factors, and ‘information quality’ with the factors that influence the quality of information and its sources. Table 1 System components and corresponding research domains System Components Information needs analysis Information sources identification Information collection Information mapping Information personalization Information provision
Corresponding Research Domains - Critical Success Factors (Rockart, 1979) - Industry Analysis (Porter, 1980) - Business Information (Lowe, 1999) - Information Seeking Process (Marchionini, 1995; Kuhlthau, 1993) - Information Quality (Huang et al., 1999) - Information Agents (Gudivada et al., 1997; Klusch, 2001) - Information Organization / Information Classification (Taylor, 1999; Rubin, 2000; Svenonious, 2000) - Information Filtering / User Modelling (Oard, 1997; Belkin and Croft, 1992) - Information Retrieval (van Rijsbergen, 1979; Baeza-Yates and Ribeiro-Neto, 1999)
Information collection from distributed sources without common coordination scheme, as is the case with sources on the Internet, is the focus of research on ‘information agents’, software entities that autonomously search for information in electronic environments. Information mapping is an organizational issue, which draws on research regarding the design of classification schemes for different information scenarios. The personalization and provision of information builds on the specification of user profiles
7 and application specific information needs which are directly linked with research on individualized ‘information filtering’ and ‘information retrieval’. The specification of system components, their relationships and their linkage with a range of related research areas, which could contribute to their organization, constitutes the conceptual framework for the development of the dynamic DYS-MIS for monitoring markets, market policies and competition in a dynamic chain and network environment. 4. Operational systems design 4.1 Principles The transfer of the conceptual framework into an operational application system involves the logical deduction of operational activities for individual system components and the specification of an operational sequence of activities, which best models the information circle described above regarding its capability and efficiency in solving the overall information problem in a certain application scenario. The deduction of activities utilizes a top-down approach which draws on the available knowledge bases linked to system components and results in a hierarchy of activities on different levels of aggregation (table 2). The sequence of activities does not mirror the logical sequence of system components but takes into account efficiency considerations derived from the - need for a repeated (continuous) activation of the information circle and - existence of a compound information interest of the network, which could build on a network initiated monitoring process independent of and prior to any enterprise induced information request, which, in turn, could then be served by its results. In this sector-based continuous monitoring system, the first group of activities needs to be performed only during the establishment of the system and in certain time intervals thereafter for an adaptation of the system’s basis to changing conditions. These activities specifically link the system to a sector’s network application scenario, which requires substantial knowledge of the application environment, a large degree of human expertise and extensive analytical efforts. It is these activities where the general knowledge bases from related research domains are not sufficient for the delineation of operational activities. The deficiencies
8 relate to both, system specific knowledge of the application environment as well as general knowledge on system organization. As the research domains have not been directly linked with the conceptualisation of information systems for dynamic supply networks, they do not provide a comprehensive basis for system organization. Our approach complements existing knowledge on system organization and provides a guideline for the identification of system specific knowledge of the application environment through additional focused research activities and the integration of comparably focused expert knowledge. Table 2 System components, hierarchy of activities and knowledge base deficiencies COMPONENTS 1
5 2
3 4
6
ACTIVITIES
SUB-ACTIVITIES
Group 1 (infrequent activation) 1. Analysis of needs Global analysis of needs of different target Specification of inform. groups distribution scheme 2. Determination and Creation of taxonomy Information implementation of Specification of filters and personalization taxonomy and filters user roles Inform. sources 3. Determination of Search for sources identification Internet sources Quality assessment Group 2 (frequent/continuous activation) 4. Monitoring of Information Access of sources collection Internet sources Gathering of documents Indexing of documents 5. Information Information Integration of documents mapping categorization into taxonomy scheme Group 3 (frequent activation upon request) Filtering and retrieval 6. Information Information Integration into enterprise provision provision inform. processes Information needs analysis
E1) X X X X X X
1) Necessity for integration of system focused additional research and expert knowledge
The second group of activities is the core of the sector-focused continuous monitoring process whereas the last activity is activated by individual enterprises’ information requests. The frequency of their activation makes their operational efficiency a crucial issue for overall system performance. However, it is also the activities of the two last groups, which have been at the centerpoint of research efforts on knowledge bases for
9 system automation. In our sequence of activities, these knowledge bases are being complemented by additional and more focused knowledge acquisition efforts during the first group of activities. A case in point are the activities linked to the system component ‘personalization of information’. During system initialisation, the personalization activity could design rules for the design of a scenario-specific computerized information filter, which is then being used as a substitute for the activity in subsequent information circles (activity 6). In the sequence of activities, the activity ‘specification of an information distribution scheme’ has a prominent position as a key transformer of information needs and priorities into system infrastructures for serving the needs. It is a crucial point where decisions are being made concerning information needs, which could be considered as ‘relevant’ for the network as a whole or for distinguished groups within the network (e.g. enterprises of a certain product supply line or at a certain stage of the supply network) and those with a predominantly individual orientation. The network relevant needs are embedded in the ensuing search process and modelled in a subject oriented taxonomy hierarchy, which groups collected information according to interests. Information interests of enterprises are being served through the taxonomy model but can be further individualized through the information filters, which identify enterprise relevant information from any of the distinguished taxonomy subject groups according to individual selection criteria. 4.2 Baseline activities
The initial analysis of market-focused information needs of distinguished target user groups builds on a well-established scientific basis. Furthermore, there is sufficient empirical evidence on the suitability of the approaches for sector-based but enterprise focused information needs analysis (e.g., Kuron, 1993) if applied by experts with ‘knowledge’ about sector problems, i.e., if combined with scenario specific expert knowledge. In a stepwise systematic analysis process, broad information topics are mapped in a hierarchical model, which splits broad information topics into narrower information indicators and further into related information items. On each level, results
10 can be captured in matrices of information needs which link information topics, indicators and information items to distinguished groups of enterprises and highlight basic synergies and differences. The transformation of the hierarchy of information needs into a service oriented system infrastructure (activity 2) involves the modelling of the taxonomy model (Fritz, 2002) and the specification of the information filters. In this two-tier information structuring process, efficiency considerations determine the shift from taxonomy modelling with its focus on collection, grouping and differentiation to filtering with its focus on selection and use. The identification and evaluation of information sources for subsequent monitoring activities (activity 3) is a complex and crucial task. The approach combines research results regarding information seeking processes and information quality assessments with application specific expert knowledge regarding a network’s information environment into a formalized two-level search process (Fritz, 2002). It builds on a multi-level catalogue of selection and assessment criteria and identifies, in a first step, relevant sources, which, in a subsequent step, are evaluated regarding their value for the supply network in terms of content and quality. However, in this phase, the level of formalization is limited. The quality of results and the efficiency of search still depends on the availability of additional, difficult to formalize expert knowledge on the dynamic information sphere of the Internet. 4.3 Automation potential and process configuration
The complexity of information requirements for the totality of enterprise actors within a dynamic network situation suggests and demands a large degree of automation to reach economic feasibility. This applies especially to activities, which are most frequently activated during the course of the information circle. They involve the - continuous monitoring of information sources (activity 4), - preparation of collected information for use (activity 5) and - guided provision of information according to actors’ requests (activity 6).
11 Those activities have been the focus of interest regarding automation opportunities from two different angles of research, environmental scanning research (e.g., Stoffels, 1994) with its interest in scanning routines for information collection from business environment, and information agent research (Klusch, 2001) with its interest in the integration of ‘intelligence’ in robots for unstructured information search in poorly structured electronic information environments (for other research directions in agent technology see, e.g., Kirn, 1999). Both lines of research build, in principle, on a systematic utilization of expert knowledge (or any other type of ‘intelligence’), which might reach from the application of certain rules for systematic analysis to sophisticated rule-based computerized expert systems, an operational first step towards really ‘intelligent’ systems as indicated for electronic information agents. Based on an evaluation of the complexity of the different tasks, the generalization potential of expert knowledge needs, the state of current information agents research and technology (Fritz, 2001a,b) and the potential efficiency benefits, the proposed dynamic monitoring system could build on automation opportunities which fully cover the continuous monitoring part of the dynamic monitoring process (figure 3).
ACTIVITIES FOR KNOWLEDGE ACQUISITION (preparation und periodical revision; require expertise)
Analysis of inform. needs of target groups Inform. distribution scheme
Implementation of taxonomy and filters
Sub-activities
Determination of internet sources
AUTOMATED ACTIVITIES (software-based execution) Source monitoring
Categorising
Inform. provision
Basic monitoring routine
Figure 3: Organization and scheduling of activities of an operational dynamic sectorbased market monitoring system (size indicating routine efforts involved)
12 For the continuous processes of monitoring the selected information sources and of categorising collected information for later use, information agents technology has reached a level of sophistication in research based selected software products, which makes it a suitable choice for implementation. In our case, information agents are supposed to regularly visit pre-defined information sources on the Internet, detect changes in the sources’ information provision, gather new pieces of information and categorise collected information with respect to the taxonomy model. They have to be supplied with guidance on how to autonomously and automatically sift through the predetermined information space within the Internet (which is provided through a knowledge base from general agent research) and a search strategy regarding source and frequency of monitoring (which needs to be provided by application specific expert knowledge). For the access of enterprises to the monitoring results, information filter technology offers a principal option for efficient personalization without human intervention. The integration of information filter technology in some of the agent-based software products supports its integration into the system. Its implementation and connection with internal processes of enterprises needs to be guided by application specific expert rules developed in the initial personalization stages of the dynamic monitoring process. 5. Process activation and system validation
The complexity and diversity of the scientific background of the dynamic monitoring system’s design is reflected in the complexity of the system’s activation in a selected application environment and the necessary efforts to maintain required quality levels and to continuously adjust to increasing information needs. As the system does not build on a comprehensive logical model or a complete basis of integrated scientific evidence of the quality and efficiency of its delivery it requires empirical validation. Furthermore, the dynamic monitoring system’s need to adapt to changing environments and changing information needs of its enterprise actors requires subsequent validation efforts regarding the quality of application oriented embedded
13 rules and actual expert involvement. However, specifics of system design allow a reduction of validation complexity in system activation and system adaptation. As each of the activities builds on its own knowledge base and generates its own and clearly identifiable results, validation may as well limit its focus on the knowledge base and especially the integrated expert knowledge of individual activities. Furthermore, through the system’s linkage with a variety of scientific research domains empirical validation can bypass some basic system elements where logical reasoning or scientific evidence are convincing enough for acceptance without any further validation initiatives. Our approach accepts this argument for the - logic of the information circle and the deducted principal sequence of activities, - principal approaches for the analysis of information needs, and - application independent ‘intelligence’ in information agents technology. The initiation of the first group of activities does not build on a strict application of integrated process rules but on a combination with some additional expert involvement. System validation has to account for this combination through a clear specification of the validation focus. 6. Empirical results and future research
Ongoing research on the design of dynamic monitoring systems for the agri-food sector do focus on three lines of research results: - the formulation and validation of the activities’ sets of process rules and especially the application oriented rules that complement results of scientific research in the initialising group of activities, - the establishment and validation of a sequence of activities which balances quality needs and efficiency requirements on system and activity level and - the design and validation of a model implementation for various product lines of the agri-food sector. The first line of research provides a knowledge base which supports the activation of the crucial initialising activities in the agri-food sector. The second line allows drawing conclusions on a best division between sector and individual information initiatives in the
14 sector and a best balance of efforts regarding the activation and implementation of various activities. The third line allows mapping a principal information infrastructure for a suitable monitoring system in the sector regarding the monitoring of markets, market policies and competition activities. First simulation studies have demonstrated the principal feasibility of the dynamic monitoring process and the individual activities. Pre-tests regarding the usability of rule sets on the identification and evaluation of Internet information sources without any additional individual expert involvement have shown satisfactory results across different product lines and different levels of the supply chain. As supported by a variety of studies related to different subsectors of the agri-food sector (see, e.g., Kuron, 1993), the aggregate information needs (information topics) derived from a sector focused critical success factors analysis with enterprises show little variation between different product lines and across different levels of the supply network. Differences develop in lower levels of the hierarchy primarily separating different levels of the supply network and different product lines. Further studies need to focus on the balance between quality, process efficiency and economic feasibility. This includes the formulation and validation of filters, the specification of the boundary between information search (as specified by the taxonomy model) and information retrieval (as specified by the filters), the size of the search domain, and the monitoring frequency. 7. Summary
In dynamic supply networks and chains with temporarily evolving supply chain partnerships in complex market situations, enterprises depend on interactions among themselves and with their business environment which match the dynamics of the network situation. Continuous awareness of developments in markets, market policies and competitor’s activities is an essential prerequisite for economic success. This paper has focused on the design of an operational dynamic and individualised sector monitoring system to assist management of network and chain actors in their scanning activities in the information environment of the Internet and to provide them
15 with continuously focused information on markets, policy and competition. It linked the discussion to the agri-food sector which closely resembles a dynamic supply network situation and provides the basis for the empirical validation of the approach. The concept requires the consideration of results from different domains of scientific research, utilizes established expert knowledge, eliminates deficiencies through own research results and integrates automation potentials offered by intelligent agent and information filter technology to assure economic feasibility. The dynamic system approach depends on a continuous validation process with various levels of validation needs in experimental or sector application settings. First studies have demonstrated the principal feasibility of the dynamic monitoring process and its individual activities. However, to assure economic sustainability in a selected application environment, future studies need to specifically focus on the balance between information quality, process efficiency and economic feasibility. References
Baeza-Yates, R., B. Ribeiro-Neto, B., 1999, Modern Information Retrieval, ACM Press, New York. Belkin, N. and B. Croft, B., 1992, Information filtering and information retrieval: Two sides of the same coin? Communications of the ACM, 35 (12), 29-38. Copacino, W.C., 1997, Supply Chain Management. The basics and beyond, St. Lucie Press, Boca Raton. Fritz, M., 2001a, Intelligent information agents: An introduction, Research Report A-01/1, University of Bonn-ILB Press, Bonn (in German) Fritz, M., 2001b, Internet-based information systems. Software evaluation – industry analysis according to PORTER, Research Report A-01/3, University of Bonn-ILB Press, Bonn (in German) Fritz, M., 2002, The Internet as source for business information: identification and quality assessment of appropriate websites, Research Report A-02/2, University of Bonn-ILB Press, Bonn (in German) Gudivada, V.N., V.V. Raghavan, W.I. Grosky and R. Kasanagottu, 1997, Information Retrieval on the world wide web, IEEE Internet Computing, 1 (5), 58-68 Huang, K.-T., Y.W. Lee and R.Y. Wang, 1999, Quality Information and Knowledge, Prentice-Hall, Upper Saddle River
16 Kirn, St., 1999, New German Priority Research Program on "Intelligent Agents and realistic Commercial Application Scenarios" [2000-2006], Workshop Notes of the 1999 United Kingdom Workshop on Multiagent Systems, Bristol, 6-7 December 1999 Klusch, M., 2001, Information Agent Technology for the Internet: A Survey, Journal on Data and Knowledge Engineering, Special Issue on Intelligent Information Integration 36 (3) Kuhlthau, C.C., 1993, Seeking Meaning. A Process Approach to Library and Information Services, Ablex Publishing, Norwood Kuron, U.H., 1993, ERP Systems in Commodity Trade, Wehle Publisher, Bonn (in German). Lowe, M., 1999, Business Information at Work, Aslib / IMI, London Marchionini, G., 1995, Information Seeking in Electronic Environments, Cambridge University Press, Cambridge Ménard, C., 1996, On Clusters, Hybrids, and Other Strange Forms: The Case of the French Poultry Industry, Journal of Institutional and Theoretical Economics (JITE), 152 (1), 154-183 Oard, D., 1997, The state of the art in text filtering, User Modeling and User-Adapted Interaction, 7 (3), 141-178 Porter, M., 1980, Competitive strategy: techniques for analysing Industries and competitors, The Free Press, New York Rijsbergen, C. van, 1979, Information Retrieval, 2nd ed., Butterworths, Boston Rubin, R.E., 2000, Foundations of library and information science, Neal-Schuman, New York Rockart, J.F., 1979, Chief executives define their own data needs. Harvard Business Review, 58, March-April, 81-93 Stoffels, J.D., 1994, Strategic issues management: a comprehensive guide to environmental scanning, Pergamon, Oxford Svenonious, E., 2000, The intellectual foundation of information organization, MIT Press, Cambridge Tan, K.C., 2002, Supply Chain Management: Practices, Concerns, and Performance Issues, The Journal of Supply Chain Management, 38 (1), 42-53 Taylor, A.G., 1999, The organization of information, Libraries Unlimited, Englewood Vorst, J.G.A.J. van der, 2000, Effective food supply chains. Generating, modelling and evaluating supply chain scenarios, Diss., Wageningen Wigand, R.T., Picot, A., Reichwald, R, 1997, Information, Organization and Management. Expanding Markets and Corporate Boundaries, John Wiley & Sons, Chichester
