A Simulation Based Aid for Complex Dynamic Decision Making

Abstract

Modern organisations are large complex systems operating in an increasingly dynamic environment and are tasked to meet its goals by adopting suitable courses of action. Deciding an appropriate course of action calls for deep understanding of various aspects of organisation such as its goals, structure, business-as-usual operational processes and business dynamics. The state-of-practice of decision-making that relies heavily on human experts is often reported as ine ective, imprecise and lacking in agility. Dissertation presented in this paper aims to develop a suitable aid that will assist decision makers to arrive at e ective decision with increased rigour, precision and agility for complex dynamic decisionmaking activities.

Keywords

• Organisational decision-making
• Simulation based decisionmaking
• Socio-technical System

-

Modern organisations constantly attempt to meet organisational goals by adopting appropriate courses of action [ 27 ]. Evaluating possible courses of action and selecting option that has the best potential to meet organisational goals are challenging task. It requires precise understanding of various aspects of an organisation such as goals, organisation structure, operational processes, historic data and the stakeholders of the organisation [ 12 ]. Comprehending relevant aspects in precise and meaningful manner to address complex dynamic decision-making (CDDM) is a challenge as it deals with large organisation with socio-technical characteristics [ 20 ], uncertainty and non-linear causality in business interactions [ 10 ], and high business dynamics [ 29 ].

The industrial practice of organisational decision-making heavily relies on human experts who are aided with primitive tools such as spreadsheets, word processors, and diagram editors. Though adequate for capturing and collating the required information, these tools o er limited analysis support if at all [ 19 ]. This limited use of technological aids in analysing information make the decisionmaking activities time-, e ort- and intellectually-intensive endeavors. Reports from leading consulting institutions such as McKinsey and Harvard Business Review [ 15, 22 ] often state this state-of-the-practice as biased, based on shortterm emotion and imprecise for current business context. The problem is more critical for CDDM kinds of decision-making problems due their inherent characteristics.

Dissertation presented in this paper attempts to improve the situation by conceptualising and developing a simulation platform for organisational decisionmaking, precisely for CDDM. In particular, the research aims to improve precision, reduce personal biases, consider short term and long term e ects, reduce the excessive burden on human experts and provide a-priori indication in shorter time window by introducing a simulation-based evaluation platform to the decision makers. The primary contributions of this research work are a) a language or meta-model to specify CDDM problems in a comprehensive manner, and b) a simulation platform for a-priori evaluation of decisions through what-if and if-what analyses. These two research contributions form the core enabling technology of an overarching solution1 that aims to address organisational decisionmaking problem using a pragmatic business facing decision-making framework wherein the practitioners can capture enterprise using problem speci c terminologies and pose their questions leading to decision-making using higher-level language abstraction.

This paper presents an overview of proposed research agenda, research progress and the plan toward the completion. The paper proceeds as follows: section 2 presents research motivation and objectives; methodology adopted for this research is presented in section 3; the research overview including the problem formulation, high-level approach and validation plan is discussed in section 4. Paper concludes with a brief summary on research progress, achievements and future plans. 2

Research Motivation and Objectives

Industry practice follows re nement-based methods such as Incremental method [ 24 ] and Carnegie Method [ 11 ] for organisational decision-making. These decisionmaking methods are essentially guided by set of decision questions that includes: What are the overall goals? Are there any dependencies between these goals? What are the course of actions to achieve them? How do these course of action impact the goals? How they di er qualitatively and quantitatively? The excessive dependency on human cognitive capability and limited utilisation of technological aids to answer these decision questions results into ine ective decisions. 1 http://www.tcs.com/research/Pages/Model-Driven-Organization.aspx Essentially, the decisions are often biased as the information syntheses are based on personal interpretation [ 15 ], limited to short term bene t due to cognitive limitation [ 22 ], and not precise enough due to lack of tool assisted analyses [ 19 ]. In contrast, the practitioners expect better decision-making approach and tool support that can improve precision, reduce personal biases, consider short term and long term e ects, reduce excessive burden on human experts and provide a-priori indication about the e cacy of decisions.

This perceived poor view of the state-of-practice of decision-making (particularly for CDDM) elicits a research question: What kinds of technological aids will help decision makers to arrive at precise, unbiased and e ective decisions? The dissertation presented in this paper focuses on technical aspect of this broad research question wherein the key objectives considered are: 1. Improve precision and e ectiveness of the decisions with appropriate machinery to evaluate them a-priori. 2. Reduce excessive dependency of human experts to understand short-term and long-term implications. 3. Improve the agility of organizational decision-making, i.e., reduce the analysis time and e ort, with automation. 3

Research Methodology

This dissertation adopts Design Science Research (DSR) methodology proposed by Hevner in [ 14 ] for conducting research activities. Primarily, it follows three DSR cycles namely relevance cycle, design cycle and rigor cycle using the ve research activities: problem statement conceptualisation, exploration of state-ofthe-art and state-of-the-practice of organisational decision-making, conceptualization of proposed approach, implementation of conceptualized approach, and research validation. The iterative execution of two activities - a) problem statement conceptualisation and b) exploration of state-of-the-art and state-of-thepractice of organisational decision-making form the relevance cycle. The problem statement conceptualisation activity considers practitioner's views and industrial reports. The exploration of state-of-the-art and state-of-the-practice of organizational decision-making uses two methods - literature review based exploration and experiment based exploration to understand the capabilities and limitations of existing approaches and tooling infrastructures in the context of CDDM. Literature review based exploration adopts evidence-based secondary studies, such as systematic mapping study (SMS), [ 25 ] and experiment based exploration uses synthetic but close to real life case studies.

The design cycle comprises three iterative activities namely conceptualization of proposed approach, implementation of conceptualized approach, and research validation. This research adopts meta-modelling approach to arrive at suitable conceptual model and language processing paradigm for implementation artefacts. The research validation is based on Arti cial and Ex-Post [ 26 ] evaluation strategy. It is Ex-Post as the evaluation to be performed after design and development of research artefact, and Arti cial as the synthetic case studies illustrating scenario from industry and academia to be considered for validating research artefacts.

Finally, the rigor cycle that establishes the connection between research outcomes and knowledge-base to be performed using meta-analyses on multiple Ex-Post evaluations from research validation activities. 4 4.1

Research Overview

Problem Statement

A conceptual representation of decisionmaking is depicted in Fig. 1. As shown in the gure, an Organisation has a set of Goals, it publishes set of key performance indicator or Measures.

The decision makers analyse, observe or predict Measures and decide appropriate courses of action or Levers incase they nd that the Goals are not achievable or not achieved. The key Fig. 1. Organisational decision-making activity of the decision-making is to select appropriate Levers for the stated Goals. It is an iterative exploration and evaluation of the available options to nd best possible option that has potential to achieve Goals. The e cacy of such exploration depends on two key factors: (i) the ability to capture relevant information about Organisation and its environment, and (ii) the ability to perform what-if and if-what analyses, e.g., what will happen in terms of Measures and Goals if speci c Lever is applied to Organisation or which Levers can leads to speci c Measures, etc.

This dissertation argues that an Organisation can be understood well by analyzing what an enterprise is, how it operates, why it is so, and who are the responsible stakeholders [ 6 ]. This hypothesis is principally aligned with the Zachman framework [ 35 ]. The enterprise architecture frameworks, such as ToGAF2, further advocate the need for a holistic view of an organisation for comprehensive understanding. Thus an ability to establish the relationships between various aspects constitutes a requirement.

Decision-making can progress either top-down or bottom-up. The former is a re nement process wherein the decision-maker begins by specifying enterprise at a coarser level of granularity by ignoring details. Bottom-up decision-making is a converse of top-down with abstraction replacing re nement. Therefore, ability to be cognizant of abstraction and re nement relationships across levels is a critical requirement on enterprise speci cation.

CDDM puts some special demands on speci cation in terms of desirable characteristics of organisation that include reactive, adaptable, modular, autonomy, intentional, compositional, uncertainty and temporal as described in Table 1. 2 https://www.opengroup.org/togaf/

Modular l Composable icaReactive hnAutonomy ce Intentional t i-oAdaptable coUncertainty STemporal s Visualisation i syMachine lanInterpretable AQuantitative

Qualitative

Intentional Speci cation Structural Speci cation Behavioural Speci cation Speci cation on stakeholders and responsible human actors Must encapsulate internal goal, structure and behaviour.

Multiple parts should be composed to a consistent whole.

Must respond appropriately to its environment Possible to produce output without any external stimulus.

Intent de nes the behaviour Adapt itself based on context and situation Precise intention and behaviour are not known a-priori.

Inde nite time-delay between an action and its response Support for visualization Models that are interpretable by machine (i.e., support for simulation/execution) Simulation based quantitative analysis

Simulation based qualitative analysis

Furthermore, industry practice of decision-making desires precise what-if and if-what analysis for a-priori indication of decision. As a result, ability to assess a decision in qualitative and quantitative manner forms the basis of analysis requirements. A-priori assessment of decisions is suggestive of simulation capability. Table 1 enumerates speci cation and analysis requirements for CDDM. 4.2

Literature Review and Experiments

The problem de nition triggers several research questions, such as: What kinds of modeling abstractions and analysis techniques are available for specifying and analysing di erent aspects of an organisation? Are they capable of supporting expected characteristics of CDDM? What are the gaps? A series of systematic mapping studies and experiments are conducted to nd comprehensive answers to these research questions.

A literature review on Enterprise Modeling (EM) literature using SMS methodology was performed to evaluate their suitability in the context of CDDM [ 5 ]. The review concluded with a critical observation that the existing EM techniques are capable of satisfying the expected requirements of CDDM described in Table 1 in parts. The review synthesis led to an exploration on multi-modelling and cosimulation environments involving multiple EM techniques to address CDDM. The exploration is conducted using two approaches: a) a literature review on multi-modelling and co-simulation environments such as DEVS [ 9 ], AA4MM [ 28 ], AnyLogic [ 8 ], and b) an experiment on multi-modelling and co-simulation approach by combining i* [ 34 ], Stock-and-Flow [ 21 ] and BPMN [ 33 ] tools. The research nding, experimental setup and experiences are presented in [ 16 ]. Both the literature review and experiment on multi-modelling and co-simulation approach have produced evidences that multi-modelling and co-simulation based approach are largely prone to intrinsic complexity [ 17 ] and accidental complexity [ 16 ]. Moreover, they lack in expressing several socio-technical characteristics such as autonomy, uncertainty, temporal behaviour and adaptability.

The inadequate support for socio-technical characteristics in EM techniques (in isolation or within multimodeling set up) opens up a scope for exploring the languages and frameworks that are based on actor model of computation [ 1 ]. A literature review using SMS methodology on actor language and frameworks discloses their suitability in the context of CDDM. Essentially actor languages (e.g., Erlang Fig. 2. Conceptual Model [ 3 ], SALSA [ 32 ], AmbientTalk [ 31 ], and Kilim [ 30 ]) and frameworks (e.g., ActorFoundry [ 4 ], Scala Actors [ 13 ], Akka [ 2 ]) are capable of specifying and analysing a range of socio-technical characteristics such as reactive, modular, autonomy, intentional, compositional and emergent behaviour. The identi ed limitations of actor languages and frameworks are: a) lack of support for the notion of time and uncertainty, b) explicit support for relevant aspects, their relationships such as abstraction and re nement relationships. 4.3

Proposed Approach

This research proposes a conceptual model to describe CDDM and conceptualizes a decision-making platform by using the concepts and technology explored in previous section. The conceptual model and overview of proposed platform de nition are described below.

Conceptual Model: An organisation can be viewed as something that responds to a set of events as it goes about achieving its stated goals. Oganisations consist of many autonomous units, organised into dynamically changing hierarchical groups, operating concurrently, and managing goals that a ect their behaviour. The structure and behaviour of an organisation are described using a set of concepts as depicted in Fig. 2.

A Unit that represents organisation is an autonomous self-contained functional unit with high coherence and low external coupling. It exposes Goals stating its intention. It interacts with environment through a set of In-Events and Out-Events. Internally it contains a Behaviour, a set of Internal Events and current and previous states of the organisation, i.e. Data and Trace.

A Unit may make use of several contained Units in order to meet the promised goals. The contained units can interact with each other to delegate their responsibilities to others; a unit can also participate in hierarchical composition structure to accomplish wider goals of the organisation. A Unit has a set of Levers and Measures where levers are transformation function and the parameters that can be used for con guration purposes, and measures are meaningful state variables that are exposed to the environment.

Conceptually, elements Unit, Event, Data, Trace and nesting capability of Unit, collectively, speci es the what aspect, Goal speci es the why aspect, Behaviour speci es the how aspect and Unit, as individual, speci es the who aspect of an organisation. Event helps to capture reactive nature, the intent is captured using Goal, modularity is achieved through Unit, autonomy is possible due to the concept of Internal Event, and composition can be speci ed using nesting relation. Also, Unit is adaptable as it can construct and reconstruct its structure; modular as it encapsulates the structure and behaviour of an organisation; intentional as it has its own goals; and compositional as it can be an assembly of Units.

A set of existing concepts are adopted to come up with Unit abstraction. Modularisation and re ective unit hierarchy are taken from fractal component models [ 7 ]. Goal-directed reactive and autonomous behaviour can be traced to agent behaviour. De ning states in terms of a type model is borrowed from UML. An event driven architecture [ 23 ] supports exible interactions between components, and the concept of intentional modelling [ 34 ] is adopted to enable speci cation of component goals.

Decision Making Platform: A high-level platform de nition is depicted in Fig. 3. Platform supports two languages namely Decision Making Speci cation Language (DMSL) and Execution Language (EL). DMSL concretises proposed conceptual model to specify the CDDM instances. It speci es the organisation, possible levers, expected measures and the environment where an organisation operates. DMSL also speci es various aspects of organisations in relatable form, the socio-technical characteristics and the various relationships such as abstraction and re nement as described in Table 1. Execution language is executable speci cation to perform what-if and if-what analyses.

DMSL uses the concepts borrowed from actor model of computation, eventdriven systems, declarative rules, goal speci cation, conventional class model, linear temporal logic and theory of uncertainty. As part of overarching research initiative, a language termed as enterprise speci cation language (ESL) is designed and prototyped by extending actor model of computation with the notion of time and uncertainty. This dissertation plan to use ESL as underlying execution machinery by supporting adequate mappings to translate DMSL to ESL. Proposed platform de nition uses primitive statistical tools to interpret and visualize simulation results of ESL. 4.4

Validation Plan

The research outcome will be validated using two synthetic near real-life casestudies: a) a case study on software service provisioning organisation, and b) case study on academic institution. The case study on software service provisioning organisation (as illustrated in [ 16 ]) focuses on improving the revenue of an organisation by exploring possible levers such as project selection, recruitment strategy and investment of tooling infrastructure. The case study on academic institution explores the possibility of improving university ranking by deciding appropriate levers such as research collaboration, teaching and research ratio, PhD student intake, sta selections, etc. 5

Conclusion

The research presented in this paper is part of an overarching research agenda of an industrial research organisation that aims to develop a business facing platform as an aid to the decision makers to arrive at precise, unbiased, evidence based decisions. It also aims to reduce the excessive human dependency and improve the agility of the decision-making process. This dissertation focuses on the technical aspect of overarching research agenda by introducing a language to specify decision-making instances and simulation platform to evaluate the e cacy of the possible decisions.

Till date, the problem statement is de ned by evaluating industrial reports and its relevance is validated though a list of publications [ 18, 17, 16, 6 ]. The research scope and contributions are identi ed using literature reviews and experiments on various options that have the potential to realise decision-making platform. The literature review on Enterprise Modelling technique is presented in [ 5 ] and experiments of multi-modelling option is presented in [ 16 ]. The hypotheses and conceptual model are presented in [ 16, 6 ]. De nition of DSML and exploration of candidate execution language are ongoing activities. As next step, this research will also focus on game theoretic approach, competition and collaboration aspects for better decision-making. A recommendation system for guided simulation using appropriate technique such as genetic algorithm to be explored in this research.

References

1. Agha , G.A. : Actors: A model of concurrent computation in distributed systems . Tech. rep., DTIC Document ( 1985 )

2. Allen , J.: E ective akka. " O 'Reilly Media , Inc." ( 2013 )

3. Armstrong , J.: Erlang - a survey of the language and its industrial applications . In: Proc. INAP . vol. 96 ( 1996 )

4. Astley , M.: The actor foundry: A java-based actor programming environment . University of Illinois at Urbana-Champaign: Open Systems Laboratory ( 1998 )

5. Barat , S. , Kulkarni , V. , Clark , T. , Barn , B. : Enterprise modeling as a decision making aid: A systematic mapping study . In: 9th IFIP WG 8.1 Working Conference on The Practice of Enterprise Modeling (PoEM) , Skovde, Sweden ( 2016 )

6. Barat , S. , Kulkarni , V. , Clark , T. , Barn , B. : A simulation-based aid for organisational decision-making . In: ICSOFT-EA 2016: 11th International Conference on Software Engineering and Applications ( 2016 )

7. Barros , T. , Ameur-Boulifa , R. , Cansado , A. , Henrio , L. , Madelaine , E.: Behavioural models for distributed fractal components . annals of telecommunications-annales des telecommunications 64(1-2) , 25 { 43 ( 2009 )

8. Borshchev , A. : The big book of simulation modeling: multimethod modeling with AnyLogic 6 . AnyLogic North America Chicago ( 2013 )

9. Camus , B. , Bourjot , C. , Chevrier , V.: Combining devs with multi-agent concepts to design and simulate multi-models of complex systems (wip) . In: Proceedings of the Symposium on Theory of Modeling & Simulation : DEVS Integrative M&S Symposium . pp. 85 { 90 ( 2015 )

10. Conrath , D.W. : Organizational decision making behavior under varying conditions of uncertainty . Management Science 13 ( 8 ), B{ 487 ( 1967 )

11. Cyert , R.M. , March , J.G. , et al.: A behavioral theory of the rm . Englewood Cli s, NJ 2 ( 1963 )

12. Daft , R.: Organization theory and design . Nelson Education ( 2012 )

13. Haller , P. , Odersky , M. : Scala actors: Unifying thread-based and event-based programming . Theoretical Computer Science 410 ( 2 ), 202 { 220 ( 2009 )

14. Hevner , A. , March , S.T., Park , J. , Ram , S. : Design science in information systems research . MIS quarterly 28(1) , 75 { 105 ( 2004 )

15. Kahneman , D. , Lovallo , D. , Sibony , O. : Before you make that big decision . Harvard business review 89(6) , 50 { 60 ( 2011 )

16. Kulkarni , V. , Barat , S. , Clark , T. , Barn , B. : Toward overcoming accidental complexity in organisational decision-making . In: Model Driven Engineering Languages and Systems (MODELS) . pp. 368 { 377 ( 2015 )

17. Kulkarni , V. , Barat , S. , Clark , T. , Barn , B. : Using simulation to address intrinsic complexity in multi-modelling of enterprises for decision making . In: Proceedings of the Conference on Summer Computer Simulation . pp. 1 { 11 ( 2015 )

18. Kulkarni , V. , Barat , S. , Clark , T. , Barn , B. : A wide-spectrum approach to modelling and analysis of organisation for machine-assisted decision-making . In: Workshop on Enterprise and Organizational Modeling and Simulation . pp. 87 { 101 ( 2015 )

19. Locke , E.: Handbook of principles of organizational behavior: Indispensable knowledge for evidence-based management . John Wiley & Sons ( 2011 )

20. McDermott , T. , Rouse , W. , Goodman , S. , Loper , M.: Multi-level modeling of complex socio-technical systems . Procedia Computer Science 16 , 1132 { 1141 ( 2013 )

21. Meadows , D.H. , Wright , D. : Thinking in systems: A primer . Chelsea Green Publishing ( 2008 )

22. Meissner , P. , Sibony , O. , Wulf , T. : Are you ready to decide? McKinsey Quarterly, April . Date Views 8 , 2016 ( 2015 )

23. Michelson , B.M. : Event-driven architecture overview . Patricia Seybold Group 2 ( 2006 )

24. Mintzberg , H. , Raisinghani , D. , Theoret , A. : The structure of unstructured decision processes . Administrative science quarterly pp. 246 { 275 ( 1976 )

25. Petersen , K. , Feldt , R. , Mujtaba , S. , Mattsson , M. : Systematic mapping studies in software engineering . In: 12th international conference on evaluation and assessment in software engineering . vol. 17 , pp. 1 { 10 . sn ( 2008 )

26. Pries-Heje , J. , Baskerville , R. , Venable , J.: Strategies for design science research evaluation . ECIS 2008 proceedings pp. 1 { 12 ( 2008 )

27. Shapira , Z. : Organizational decision making . Cambridge University Press ( 2002 )

28. Siebert , J. , Ciarletta , L. , Chevrier , V.: Agents and artefacts for multiple models co-evolution: building complex system simulation as a set of interacting models . In: 9th International Conference on Autonomous Agents and Multiagent Systems: volume 1- Volume 1. pp. 509 { 516 . International Foundation for Autonomous Agents and Multiagent Systems ( 2010 )

29. Sipp , C.M. , Elias , C. : Real Options and Strategic Technology Venturing: A New Paradigm in Decision Making , vol. 31 . Springer Science & Business Media ( 2012 )

30. Srinivasan , S. , Mycroft , A. : Kilim: Isolation-typed actors for java . In: European Conference on Object-Oriented Programming . pp. 104 { 128 . Springer ( 2008 )

31. Van Cutsem, T. , Mostinckx , S. , Boix , E.G. , Dedecker , J. , De Meuter , W.: Ambienttalk: object-oriented event-driven programming in mobile ad hoc networks . In: Chilean Society of Computer Science , 2007 . SCCC'07. XXVI International Conference of the. pp. 3 { 12 . IEEE ( 2007 )

32. Varela , C. , Agha , G. : Programming dynamically recon gurable open systems with salsa . ACM SIGPLAN Notices 36 ( 12 ), 20 { 34 ( 2001 )

33. White , S.A. : BPMN modeling and reference guide: understanding and using BPMN. Future Strategies Inc . ( 2008 )

34. Yu , E. , Strohmaier , M. , Deng , X. : Exploring intentional modeling and analysis for enterprise architecture . 10th IEEE International Enterprise Distributed Object Computing Conference Workshops ( 2006 )

35. Zachman , J. , et al.: A framework for information systems architecture . IBM systems journal 26(3) , 276 { 292 ( 1987 )