Outlining a Model of an Intelligent Decision Support System Based on Multi Agents

Performance optimization has become a necessity for the survival of enterprises as emerging technologies continue to impact them. To achieve this goal, decision making, a complex process which depends on big data and human issues, must be effective. As enterprises are being subjected to a multi-faceted pressure, they must ensure the optimization of their performance. This article investigates an intelligent decision support system (IDSS) based on multi agent systems (MAS). Our contribution consists in developing an intelligent model with an IDSS MAS approach that can detect and evaluate changes in both the external environment and the enterprise itself. This model is an adequate management tool for optimal and sustainable performance and offers real-time analytical, prospecting and optimization methods. Keywords-decision making; performance; value creation; MAS; IDSS


INTRODUCTION
Strategic decisions are the result of a complex process and they often refer to the complexity degree of decision-making issues and the environment in which an enterprise/company operates.For authors in [1], it refers to important decisions, in terms of actions taken to resources committed and jurisprudence.According to [2], there are decisions that especially focus on external environment.A decision is not always easy to take and in order to overcome the unexpected risks to manage, leaders seek to always be more effective when it comes to decision support tools usage.According to [3] decision support is considered as a science based on three assumptions: (a) postulate of reality: to have information and especially knowledge, (b) postulate of the decision-makers: act as a power actor according to the objectives set and the resources available and (c) postulate of the optimum i.e. optimize the decision.This basis principle is conditioned, in general, by various factors including environment changes and technological growth.Nowadays with the digital transformation in the field of information systems, no enterprise can escape the use of new technologies that greatly facilitate data processing and sharing.In fact, more and more specialized software offers the possibility of carefully analyzing the data for a reliable result that will allow a quick and efficient decision-making.Virtualization has become, therefore, an important tool enabling companies to be more responsive thanks to the different production and management applications.Thus, artificial intelligence has no privacy for companies because it has integrated most of their functions and has become a means of data sharing, performance measurement and optimal decision-making thanks to the combination of human and artificial intelligence.Indeed, in a highly competitive environment, enterprises seek to meet market demands and are constantly in need of solutions that enable them to be effectively reactive to the environment for greater competitiveness.The distributed decision process is an effective means of sharing, prevention and decision-making.
The proposed model, an intelligent decision support system (IDSS) based on Multi Agent Systems (MAS), is part of computer intelligence and therefore business intelligence (BI).It would enable the enterprise to collect, model and report data to provide decision support for driving and optimizing decision-making.

A. Performance
According to [4], performance is an encrypted result from the perspective of classification (in relation to oneself, improve performance / or compared to others).Long-term performance has been reduced to its financial dimension.Thus, a successful company is one that must reach the desired, by the shareholders, profitability and/or generate a certain level of profit, or hold market share that preserves its sustainability [5].Performance is the ability of an enterprise to exploit its environment in the acquisition of scarce resources essential to its operation [6].Some authors equate performance with efficiency, ability or competitiveness, productivity, success and excellence [7].The traditional definition of performance derives through a composition of efficiency and effectiveness [8].According to [9], effectiveness is evaluated in relation with specific objectives of the company and the degree of result achievement and is likely to be attached to the notion of mission as it is a measurable result to be achieved.Efficiency can be defined as the sum of outputs achieved (results achieved) by inputs units (means) engaged.The concept of performance can be defined in different ways.It means completion and it is also treated as competitiveness, effectiveness, efficiency and productivity.Commonly, the four other meanings are: • The capacity of the potential of the natural or legal person's "skills and strategy".
• The processes used to achieve the desired results, "tools".
• The results of actions measured against an endogenous or exogenous reference.
• The quantitative and qualitative indicators and success defined objectives for example satisfaction, action plan, cost and time.
Moreover, the performance's dimensions define four complementary levels related to each other which a manager can follow to set a good strategy: financial results, customer satisfaction, efficient internal processes and organizational system.With the help of an expert system, this performance can be comprehensive and will ensure effective management in all situations.We can deduce that performance occupies a very important place in enterprises.It is their main ambition and also the key business process.In [10], performance is the ability to achieve a minimum of meeting the expectations of strategic customers and optimal benefit.These are the three types of performance: • Organizational: A structure seeking to minimize the transactions.
• Social: Social effects of structure's activity are such that have ways of preventing and managing conflicts and they motivate and involve more staff.
• Economic or financial: Structure achieves profitability.
The challenge for enterprises is to improve and optimize their performance through effective decisions.Thus, the implementation of a management decision system is a value creation lever that allows them to achieve effectiveness and efficiency in the actions.

B. The Intelligent DSS
Called "System Computerized Decision Support" (DSS: Decision Support System) or "Interactive System Decision Support" or "System of Intelligent Decision Support", IDSS is an integrated computer system, designed especially for decision-making and especially aimed at company leaders.
IDSS is one of the management information system elements.It differs from the information system, since its main function is to provide not only information, but the analytical tools for decision making.It consists of programs, one or more databases, internal or external, and a knowledge database.It works with a language and a modeling program that enables leaders to study different assumptions for planning and evaluating the consequences.One of the first introductions of management decision systems (MDS) or DSS was in [11].These systems provide decision makers with an aid in the decision-making even in complex and unstructured situations.They became effective at the end of the 70s when the various operational decision support tools have emerged and took the name of Interactive DSS.The intelligent DSS is a computer system to help decision makers that deal with semi-structured problems [12][13][14][15].According to [16], a DSS is an automated information system, interactive, flexible, adaptable and specifically developed to help solve an unstructured decision problem and improve decision making.Powered by potentially all financial applications, commercial and administrative, of the enterprise, IDSS is a tool of observation and description that will enable managers to monitor activity, detect and alert management reshape the strategy to the needs of the company and its environment.However, it provides no explanation or commentary but using dashboards, reporting, and a MAS approach, the data is easily interpreted, processed and shared.

C. Multi Agent Systems (MAS)
Multi agent systems have been the subject of some studies and have been treated differently [17][18][19].The concept of fuzzy logic was introduced in [20], not only as a control methodology but also as a way to process data based on authorizing the use of membership in a small group instead of making use of membership in a cluster group.In the field of artificial intelligence, MAS reduce the complexity of solving a problem by dividing the necessary knowledge into subsets, combining an independent intelligent agent to each of these subsets and coordinating the activity of these agents [21].A multi agent system is a system consisting of agents.These agents each in its own turn, have internal interactions and are also related to the external environment.An agent is an autonomous computer system or program capable of carrying out independent actions.[20].Agents need to cooperate with each other, to have harmony amongst themselves and to carry out interactive conversation in order to have successful internal communication.A multi agent system or distributed artificial intelligence is, therefore, a set of IT processes composed of several agents sharing resources and communicating with each other [22].This communication is provided by different protocols and message types.Thus, the mechanisms of the agents' decisions are related to perceptions, representations and actions by all agents that cooperate to achieve a common task or a negotiation between those whose interests are different [23].In [24], an agent is a "computational entity" as a computer program or a robot, which can be seen as perceiving and acting independently on its environment.This is an automatic autonomy since its behavior depends on the program and experience.Development and implementation of DSS continues to hold great promise for improving decision maker's effectiveness [25].The combination of these two processes

Methodology
We consider ents and trans ult of these in ective decision ans.These K quests the ETL nsformation of t will be use orts using the the DWH agen fore informing ld his learnin ponse from t mmarizing the ntral agent will l be exploite propriate profil m with an appr the contributi ow the DWH ferent characte Disscussion IT solutions h ngineering or provement ap mits.These are ocess developm h repository an ers, our IDSS lection and pr