cultural
The risks Factors of Culture
5 October, 2016
ar2
An Information System to Support the Decision Making Process in Supply Chain Management
5 October, 2016

Decision making support in supply chains through modified petri nets and case-based reasoning

ar

Decision making support in Supply Chains Through Modified Petri Nets and Case-Based Reasoning

Dr. Inty Saez Mosquera, Inty.Saez@atlantisuniversity
Dr. Jorge Marx Gómez, jorge.marx.gomez@uni-oldenburg.de
Dr. Gilberto Hernández Pérez, ghdez@uclv.edu.cu

Abstract
From the organizational point of view, one of the main problems in the administration of the Supply Chains is the strategic alignment of all the members of the chain. This difficulty, in most of the cases, is determined by the ability of the partners to share strategies decisions in a logistical administration environment.
Independent of the context of the decisional problems that the members face, in the context of the logistical administration of the chain, it is possible to identify the common patterns. Generally, the persons in charge of making the decisions use their “memories” of their behavior in previous situations to create a new behavior that is tailored to the new conditions. We propose a general approach to modeling problem formulation and solution for these problems to identify the common patterns aforementioned, integrate the resulting knowledge and share it across the Supply Chain.

Keywords: SCM, decision-making problem structuring, Petri Net, CBR, knowledge sharing and integration

INTRODUCTION
From the organizational point of view, one of the main problems in the administration of the Supply Chains is the strategic alignment of all the members of the chain [1]. This difficulty, in most of the cases, is determined by the ability of the partners to share strategies decisions in a logistical administration environment [2,3].

Independent of the context of the decisional problems that the members face, in context of the logistical administration of the chain, it is possible to identify the common patterns [4,5]. These common patterns, that are shared, would constitute the base of a corporate benchmark running the length and width of the chain whose fundamental comparative indicator would be in fact, the effectiveness of the decisions made.

From this perspective, the problem of alignment of the strategic administration of the organizations with its informative strategy appears to be an essential key element. However, it could be necessary to formulate the following question: what type of information is necessary and must be available to reach these objectives? The use of the term of information is polisemic [6]. However, in the context that it is being used, we can respond to the question arguing that of the patterns mentioned before, the nature and content of the information should be shared among the members of the chain.

It is possible to sustain the hypothesis presented up to this point, based on the studies carried out by Nutt (1993, 1999) and presented by Corner [7] with a total of about 163 and 340 operative and strategic decisions respectively. Emphasizing the verification of the fact that at the time of making the decision the persons in charge would search their own memory for similar situations, learn and in each new interaction would incorporate this new knowledge. Generally, the persons in charge of making the decisions use their “memories” of their behavior in previous situations to create a new behavior that is tailored to the new conditions. Proceeding like this, the CBR (Case Based Reasoning) approach seems to be an alternative to support this behavior base on the skill of this tools to learn from previous case [8].

The ideas described earlier already have references in the scientific community, e.g. the proposal of the authors Biswas and Narahari [2] using the Object Oriented Modeling (OOM) paradigm; they propose a workbench for the modeling and evaluation of decisions in the supply chains context.

For the modeling of the patterns, modified Petri Nets are proposed that allow the modeling of fuzzy attributes. This demand is necessary since some situations can have subjective evaluation conditions (as in the case of attributes that are expressed in: “high”, “mid” and “low”). The methodology presented by KORIEM [9] will be used since it incorporates mechanisms of structural verification for the created networks and semantics, whenever the same semantic foundation base is shared. We propose the use of SCOR (Supply Chain Operation References) as common semantic foundation. The general approach is given in the figure 1.

ac

Figure 1. Conceptual map of general approach.

However, based on Thompson (1967) framework for decision making, organizations usually have a common repository of answer for decision problem. When one particular situation luck like previous situation (based on some common characteristics) managers and decision makers can identify and try to reuse old solution in new context. There are many different way to structuring decision problem (see [7] for more details) but there is many advantage when structuring decision problem approach are based on cause and effect relationship [10,11]. Petri net registered many use in many different contexts and this network could be interpreted as token game in simulation environments. Decision making problem could be describe by enumerations of all relevant cause-effect relationship using SCOR as common semantic meaning [12,13]. Goldratt propose the actually tree methodology for modeling the current state of any problem based on cause-effect relationships. Reading all theses relationship like rule (if .. then ..) is possible translate tree into particular Petri Net (Modified Fuzzy Petri Net – hereafter MFPN – described by Koriem in [9]).

This kind of Petri Net could be useful for modeling decision making problem in terms of full enumerations of all cause-effect relationship discovering in the context of the current situation.  In order to describe the problem, decision maker don’t need to know the actual values of the indicators (KPI indicator at each level of supply chain in terms of SCOR models) to defined cause-effect relationship. Decision maker’s only need to know which indicator have or haven’t a relation with other indicators in the context of the current situation under study. Koriem [9] also provided a methodology to verify all knowledge express in Petri Net. Hence, decision maker are free to think in the same way that he or she thinking when try to formulate decision problem and without any further efforts to explore a lot of data in order to established the actual value of indicators. This approach splits traditional way to defined problem into two different steps: first of all identify the cause and effect relationship between indicators and after find the values of indicators; providing much freedom to make the definition of decision problem in more dynamically way [7,14]. Figure 2 show how relate SCOR models, Goldratt reality tree, rules descriptions and MFPN.

The model of decision problem (describe by cause-effect relationship in the tree form) could be use as case base for CBR that assist decision making problem formulation. Using the inference capabilities of Jena framework (available at: http://www.jena.sourceforge.net) all graph store for previous decision making problem could be use as a common answer repository for organization. Better than, the combination of Jena inference capabilities and knowledge verification of Koriem methodology is possible to implements a general CBR system that assists decision making process in two different ways: by providing a mechanism to identify the common symbol describe by Thompson (1967) and to suggest possible solution due to decision problem solution could be describe in the same way that problem definition by using Goldratt future reality tree [15].

ac2

Figure 2. SCOR model and MFPN theoretical framework

Figure 3 present the general vision of proposal approach and describe how all element aforementioned work together.

ac3

Figure 3. General approach framework

CONCLUSION AND OUTLOOK

The combinations of MFPN, CBR and SCOR model – express in RDF schema – introduce a new way to assists decision making process. The use of inference capability of Jena framework enlarges the traditional boundaries of CBR mechanism and produce direct benefits to aids decision making process within organization and between organizations in the context of SC management. Koriem methodology provides a well format structure for Petri Net and knowledge verification based on specific constrains for problem context.  Petri Net enlarges the opportunities to validate problem definition by using simulations (based on token games) and to assess how strong the cause-effect relationships in the study situation context are. In the context of SC managements at strategic decision level this capabilities means the opportunities to reduce uncertainty on decision making due to the strong cause-effect assessment acting as measurement of which’s performance indictors are better to predict the whole system behavior in future scenarios.

REFERENCES

  1. McCormack, K. and K. Kasper, The extended supply chain: The statistical study. Benchmarking: An International Journal, 2002. Volume 9(Number 2). pp. 133.
  2. Biswas, S. and Y. Narahari (2003) Object Oriented modeling and decision support for Supply Chains. Electronic Article.
  3. Narahari, Y. and S. Biswas (2001) Supply Chain Management: Modelling and decision making. Electronic Article. A framework for Supply Chain Management and decision making models
  4. Geerts, G.L., W.E. McCarthy, and A. Andersen (2002) Modeling Business Enterprises as Value-Added Process Hierarchies with Resource-Event-Agent Object Templates. Electronic Article.
  5. Arnott, D.R. (1998) A Framework for Understanding Decision Support Systems Evolution. Electronic Article.
  6. Capurro, R., The concept of information, in Draft. 2002: http://www.db.dk/bh/publuk.htm. págs. 1.
  7. Corner, J., J. Buchanan, and M. Heing, Dynamic decision problem structuring. 2001, University of Waikato, Tel Aviv University. pp. 1.
  8. Summers, J.D., B.M. McLaren, and D.W. Aha. Towards Applying Case-Based Reasoning to Composable Behavior Modeling. In Conference on Behavior Representation (BRIMS). 2004. Virginia.
  9. Koriem, S.M., A Fuzzy Petri Net Tools for modelling and verification of knowledge – based systems. The Computer Journal, 2000. Vol. 43. No. 3. 2000. pp. 206
  10. Perales, J.C., A. Catena, and A. Maldonado. Aprendizaje de relaciones de contingencia y causalidad: Hacia un análisis integral del aprendizaje causal desde una perspectiva computacional. 2001. Departamento de Psicologia Experimental. : Universidad de Granada. pp:1-62.
  11. Palmero, M.L.R. La teoría del aprendizaje significativo. In Conference on Concept Mapping. 2004. Pamplona, Spain.
  12. Domb, E. and H.W. Dettmer. Breakthrough Innovation In Conflict Resolution. Marrying TRIZ and the Thinking Process. Series, Breakthrough Innovation In Conflict Resolution. Marrying TRIZ and the Thinking Process, 2006 [cited 2006]; Available from: http://www.triz-journal.com/archives/1999/05/b/index.htm. Last visited: 16.11.2006.
  13. Goldratt, E., Visión viable. 2004, Grupo TOC Perú: Paris. pp. 1.
  14. Angerhofer, B.J. and M.C. Angelides. System dynamics modelling in Supply Chain Management: research review. In Winter Simulation 2000. Berlin.
  15. Goldratt, E., No es cuestión de suerte, ed. D.d. Santos. 1999, España: Diaz de Santos S.A. pp. 1.