Cognitive Decision-Making through an Intelligent Database Agent for Predictive Analysis

SHIVANI A. TRIVEDf and REBAKAH JOBDAS

Faculty of Computer Science, Kadi Sarva Vishwavidyalaya, Gandhinagar, Gujarat 382024, India

'Corresponding author. E-mail: This email address is being protected from spam bots, you need Javascript enabled to view it , This email address is being protected from spam bots, you need Javascript enabled to view it

ABSTRACT

Cognitive decision-making is an intellectual process aimed at the selection of a course of action among various choices. Information technology applications that support decision-making processes had an evolutionary transformation of spreadsheet software to the complex decision-support systems. The cognitive process of decision-making is being practiced through various intelligent information systems. The intelligent decision support system is being used by decision makers to make smart decisions in real-time dynamic environment based on intelligent data. To achieve intelligent data, it is needed to transform existing relational data and big data into agent-oriented data and subsequent design of data model to be in time to the intelligent decision support system. For such system, the intelligent database agent (IDA) is designed to generate the analytical data for cognitive decision-making. The intelligent database agent is based on predictive data analytics. In any organization, predictive models utilize the sample which is found in historical and real-time transactional data to identify threats and prospects. IDA confines associations between many aspects to permit evaluation of threat or prospective associated with a particular set of conditions, steering cognitive decision-making. The development phases for the intelligent database agent is demonstrated and implemented to retrieve the information for cognitive decision-making in the proposed framework. The proposed framework is evaluated based on the complexity of verities of data, data query complexity, i-DSS query Execution Time, memory block used, volume of data, level of aggregation, reusability, flexibility, scalability and manageability are taken into the consideration for justification.

INTRODUCTION

In this chapter, the intelligent database agent (IDA) is designed to generate analytical data for cognitive decision-making. The IDA is based on predictive data analytics. The predictive analysis consists of a number of statistical techniques from data mining, predictive modeling, and machine learning, which analyze current and historical facts to make predictions about future or uncertain events, hi any organization, predictive models utilize the sample that is found in historical and real-time transactional data to identify threats and prospects. The IDA confines associations among many aspects to permit evaluation of threat or prospective associated with a particular set of conditions, steering cognitive decision-making. The defining pragmatic effect of these technical methods is that predictive analytics provides a probability for each individual decision variables, for example, client, worker, healthcare patient, merchandise, automobile, part, appliance, or other executive unit in order to decide, notify, or influence executive processes that can be relevant across a large number of individuals, such as in digital marketing, marketing, credit risk assessment, fraud detection, manufacturing, healthcare sector, and administrative operations. An intelligent database architecture is proposed in this chapter. The proposed intelligent database architecture is based on three layers. The first layer is user directive and intelligent acquisition. This layer acquires decision-making directives from the user. Decision makers give input to the in-data intelligence, which specifies the future needs for decisionmaking. A user directive also specifies the processing methods of that data. Data aggregation, data filtration, data integration, and other processing methods can be identified in general as a resource to generate meaningful information from data. The second layer is inherit intelligence transmission to database. This layer transforms the user directive data, the method in the specified methods defined by the user directive intelligent acquisition. In this layer, the database environment can be agent oriented, agent and operation database environment, multidimensional database environment, and distributed unstructured data environment to set up decision-making. The third layer is the in-data intelligent database. This layer embeds the bits of intelligence as in-data intelligence, that is, agent in the database. This in-data intelligence “Agent” includes base data, the method to analyze that data as needful for decision-making by user’s direction, and this in-data defined by two previous layers as an agent, in which data, method, and resource are defined for decision-making. Furthermore, in the proposed framework, development phases for the IDA are demonstrated and implemented to retrieve the information for cognitive decision-making. The proposed framework is evaluated based on the complexity of varieties of data; data query complexity, intelligent decision support system (iDSS) query execution time, memory block used, the volume of data, level of aggregation, reusability, flexibility, scalability, and manageability are taken into consideration for justification. Cognitive is the ability to retain knowledge, memory storing, conscious thinking, problem solving, and decision-making.

3.1.1 COGNITIVE DECISION-MAKING AND IDSS

Cognitive decision-making is an intellectual process aimed at selecting a course of action among various choices. Information technology applications that support decision-making processes, along with problem-solving activities, have increased and progressed over the past few decades. During the era ranging from the 1970s to 1990s, these applications had an evolutionary transformation of spreadsheet software, which was supporting simple applications to the complex decision support systems (DSSs) incorporating the use of statistical and optimization models. These systems further enriched with components of artificial intelligence to thread intelligence up to a certain level until the current era. The cognitive process of decision-making practiced through various intelligent information systems such as cognitive informatics, intelligent agent systems, expert systems, and iDSSs. Regardless of whether there is an action or opinion, eveiy decision-making process produces an ultimate selection. This growth led to the development of varieties of DSSs with different tags such as management information systems, intelligent information systems, expert systems, management support systems, and knowledge-based systems. The iDSS is being used by decision makers to make smart decisions in a real-time dynamic environment. The data realized as a precious asset for intelligent decision-making. However, these raw data are rarely beneficial as their value depends on a user”s ability to extract knowledge useful for decision support. The common challenge faced by the user in the current era of cognitive decision-making is collecting diversified rapid generation of data from various sources such as social media, sensors, computer systems, and devices. Such diversified sources generate data with characteristics such as velocity, volume, value, variety, and veracity leading data to the big data. Apart from the management concern, the efficiency and deficiency of the cognitive decision-making process depend on the techniques of data analytics.

3.1.2 INTELLIGENT DATABASES SOLUTIONS FOR COGNITIVE DECISION-MAKING USING DSS

An intelligent information retrieval depends upon the collection methods of data, storage methods, and data analytics process. To deal with the above challenges, there is a need of rigorous analytical processing on the data to support timely, intelligent decision-making through the processing technologies such as data mining, analytical tools, MapReduce, Hadoop, data warehousing, online analytical processing technologies, web-based solution, ontology-based system, data warehousing and data mining techniques, OLAP, MOLAP, etc. The iDSS is an interactive software-based system intended to help decision makers to compile useful information from a combination of raw data, documents, and personal knowledge to identify and solve problems and make decisions. The architecture of the iDSS has three fundamental components: the database (or knowledge base or big data platform), the model (i.e., decision context and user criteria), and the user (i.e., user interface). The iDSS is to enhance its capability for data management and subsequent access for the growing need of improved framework, intended to focus on modeling of data using the agent paradigm to enhance the characteristics of data as “intelligent data” in the database for accelerating information retrieval, intelligence, adaptability, system integration, and scalability of the DSS. It is needed to transform existing relational data and big data into agent-oriented data and subsequent design of the data model to be in tune with the iDSS.

LITERATURE REVIEW

Literature survey is conducted in two areas to know the trends of intelligent DSS framework development. One is cognitive decision-making, iDSS, and database solutions for an iDSS.

It has acknowledged that iDSSs are the need for cognitive decisionmaking environment in an area such as business, health, agr icultural, applied science, and biology, as well as in education [1-4]. DSSs are based on IT infrastructures to translate a wealth of information into tangible and lucrative results by way of collecting, maintaining, processing, and analyzing a large amount of complex data [6-11].

The multiagent-oriented paradigm provides cognitive decision-making in the area such as the stock market to manage portfolio and the risk management system based on preferences in the opinion of group decision-making [7,11]. A grouped DSS with an agent-oriented paradigm using frizzy clustering and the analytical hierarchy process is for determination of important data used for asynchronous processing of data for multicriteria decision. A theory of managing data allocation in a distributed database system through mobile intelligent agents is implemented in many autonomous systems. As it is indicated that intelligent agents can leam from the system and provide some statistics about the mean value of the query occurrence on specific site, representing a percentage .r of the cost can change the values to a variable to improve the response. Their research work is targeted to be implemented to improve the response of different databases on the different platform using Java Database Connectivity. An integration model for Global Resource Information Database (GRID) and multiagent system (MAS) using graphical description language called Agent Grid Integration Language, hi this, the contributors have highlighted combining with the bottom-up vision of sendee GRID and top-down approach MAS. To enhance the concept of sendee into the perspective of dynamic sendee generation [ 12,13], an agent-assisted DSS is suggested to be useful for a high degree of cooperative problem-solving capability [14]. Using agent-oriented system development techniques, one can achieve intelligent acquisition, modeling, and delivery to enhance the information system characteristics such as intelligence, adaptability, system integration, and scalability [15,16]. Additionally, it enhances the two characteristics in the DSS, that is, interoperability and reuse of underlying data files hi a heterogeneous database system using OMT and MaSE [17]. To generate learning the weighted value in the form of attributes, methods, and control mechanism hi the database of case-based reasoning in agent-oriented DSS [18]. The database management system is a primary component to cany up to date and comprehensive data for the iDSS [19]. The JADE platform and the resource allocation problem in the multiagent environment can be solved using an integrated system to present the role of the database [16,20,21]. In the overall performance of the database, read optimized Relational Database Management System (RDBMS) that contract with other systems, which are write optimized. In this concept, query execution using column representation of data rather than row as column-oriented Database Management System (DBMS) to enhance the transaction performance is demonstrated in [22]. With the use of agent technology, the database can be made self-managing, maintainable, and at optimum performance. This technology replaces human database administrators (DBAs) with intelligent agents for DBA activities. For this, the contributors have implemented the prototype using JADE and Oracle 8.1.7. It further implemented agent-based cloud services to enhance the capability, persistence, and efficiency in the cloud computing environment [20-23]. Data warehouse, knowledgebase system, and other database systems are used as backend for a DSS for this database community that used columnar database SciDB and its query language SciQL [10]. According to [10] and [24], performance of the DSS with DW and online analytical processing (OLAP) does not work effectively as traditional online transactional processing (OLTP). There are three basic components: rows and columns, predicate logic, and fixed schemes. If anyone is missing or not appropriate, then selection of traditional RDBMS is not a good choice. As specified in [25] and [26], the RDBMS has limitation in handling complex structured data, and there is need to provide a framework to design a robust database for data collection and manipulation in the DSS. Furthermore, the database framework also encompasses the integration and interoperability with the existing system [27]. In the DSS framework, the use of data warehouse, data marts, and OLAP provides the highest level of functionality and decision support with historical data to meet with business intelligence, competitive intelligence, and knowledge management along with challenges such as unstructured data, business metadata, and metadata [10,14]. The database management system plays a vital role in the DSS [26]. Complex database queries for the DSS accelerated with affordable hardware architecture with large-scale multiprocessors such as GPUs, CUDA, CellBE, and Cell SDK [26]. One of the most important features of database technology is “Trigger,” which is used to implement reactivity in database applications [28,29]. The column store database is used as an information retrieval tool with the custom index structure and the query evaluation algorithm [30]. The architecture of the traditional DSS [31 ] is closed and causes various problems such as lack of compatibility, adaptability, scalability, cost ineffectiveness, and accuracy of outcomes. The analytical study findings lead to adopting the agent-object relationship (AOR) model to achieve the stated characteristics in DSSs. Software modeling techniques such as entity relationship diagram, data flow diagram, and Unified Modeling Language are used for DSS framework modeling. Faizal et al. [32] and Taveter and Wagner [33] have worked for developing an enterprise model based on the AOR. They have defined business rules at the business level as integrity constraint, derivation rule, reaction rule, and deontic assignment. Furthermore, the business process is redefined as a social interaction process to do business. The need for deontic logic is also addressed to give importance to define in the form of the agent. In the formulating agent, the important term “RESOURCE” is introduced as they are subsumed, together with the material, information, and product. The challenges in this research are how to implement a broader view and more cognitive stance, by proposing to consider not only constrained, derivation, and reaction, but also deontic assignment rules. The agent-oriented paradigm adopted for high-level abstraction facilitates the conceptual and technical integration of communication and interaction with established information system technology [34,35]. It is mentioned in [34] that today’s information system technology is based on the metaphors of data management and data flow. Moreover, it is under pressure to adopt concepts and techniques from the highly successful object-oriented paradigm, agent orientation. It is a need to transfer data from the relational database to the object-oriented database, from the object-oriented database to the XML database, and from the object-oriented database to the agent-oriented database (AODB). For global integrity with reduction of run-time overhead provides a mapping between tuples, object, and XML schema [15,17,25,37-39]. Many researchers are working on the transformation of the relational database to the object-oriented schema. Moreover, they are working on databases to get the advantages of object-oriented databases for managing complex data structure, better storage, and flexible and expandable data design [18,40-44]. Extension of query rewriting in OODB is the solution to overcome relational model limitation [32,42,45]. Study of the current market and historical review suggests the use of NoSQL database for various types of data structure and complex data [46]. The query caching algorithm is available in query optimization techniques such as pipelining, parallel execution, partitioning, indexes, and materialized view to increasing query performance, and hits are not containing the cache mechanism in OODB [47].

 
Source
< Prev   CONTENTS   Source   Next >