Elsevier

Decision Support Systems

Volume 44, Issue 3, February 2008, Pages 689-709
Decision Support Systems

ODDM: A framework for modelbases

https://doi.org/10.1016/j.dss.2007.09.005Get rights and content

Abstract

The Web extends today's modelbases—an important DSS component—by enabling their sharing and reusing over the information network. A modelbase with a formal, precise, and expressive framework will enhance the interoperation and benefit machine–human as well as machine–machine interactions. Therefore, this paper proposes a Web-based-compatible framework for modelbases, namely ODDM. “Model Ontology” and “Model Schema” are main components of the framework. Both components seamlessly interoperate by means of OWL Declarative Description (ODD). The classification of existing modelbase frameworks is presented. Requirements for the framework are identified and the proposed framework's conformance to the requirements is also discussed.

Introduction

A Decision Support System (DSS) and its components have been impacted by Web technologies and new perspectives of DSS users. The Web has been a widely accepted information infrastructure and Web technologies have become gadgets for both server- and client-side computation, particularly the technologies that are based on eXtensible Markup Language (XML). A DSS has to inevitably exploit Web technologies to improve its accessibility and widespread employment [7], [30], [44]. Perspectives of DSS users have altered from decision models—quantitative models—as ad hoc program subroutines to decision models as services [8], [9], [50]. In short, users prefer to access a convenient decision support application and use a decision model over the information network instead of owning a copy of it. They realize that this will help overcome problems of hardware and software configuration while at the same time, eliminate the duplication of effort needed to develop similar decision models. Due to those key factors, the so-called Web-based DSS [8] has gained ground, and decision support applications and their components are being more widely shared and reused. Here, a Web-based DSS is seen as a computerized system that uses Web-based technologies and architectures as an infrastructure for decision-making. The information for decision-making can then be distributed with widespread access (either internally or externally) via Web-browsers.

Similar to an ordinary DSS, a Web-based DSS mainly comprises a database and a modelbase as its important substances [44]. The former performs data organization and retrieval using database queries; the latter stores a collection of decision models. This paper refers to the term “decision model” as a quantitative model used in Management Science and Operations Research (MS/OR) such as optimization and financial models. Like a database, a modelbase also needs a data model. The data model of a modelbase is a framework that the modelbase uses to represent and describe its decision models as well as their relationships and constraints.

Various frameworks for modelbases have been proposed by a number of authors. However, they have limitations in exploiting the opportunities the Internet provides and in offering an immediate adaptation to a Web-based environment. Notable limitations are those related to Web-deployable capability, advertisement and discovery, interoperability, expressiveness, model consolidation, and model execution.

Some existing frameworks do not provide adequate Web-deployable capability (e.g., [11], [13], [15], [33]). Such frameworks not only complicate Web deployment but also are incompatible with Web-based technologies that grow rapidly, e.g., XML and Web Services. Model advertisement and model discovery on the Web may be discouraged by frameworks that allow announcing merely simple information and provide searching by simple keywords (e.g., [9], [11], [20], [21]). Consequently, a suitable decision model will possibly not be elicited if its description is syntactically different from words or phrases in a search. Frameworks that lack interoperability demand Web-based DSS applications perform excessive interpretation and mapping effort when combining or exchanging different information defined by different enterprises. The expressiveness is limited in frameworks that are not sufficiently flexible to formally represent certain information such as integrity constraints, mathematical notations, and relationships among decision models. Performing model joining and integration—linking or combining multiple models when only one model cannot serve the need—is awkward if there is insufficient description of decision models or an inadequate conflict solution mechanism. Lastly, a decision model may not be executed on the Web to obtain a solution result (e.g., [13], [36]), as some frameworks have no supportive computation mechanism.

This paper proposes a generic framework for modelbases called OWL Declarative Description for Modelbases (ODDM). The framework aims to flexibly represent various types of decision models. ODDM is suitable for assimilating into the Web-based infrastructure and is ready to be enhanced. It is based on OWL (Web Ontology Language) which is the ontology language recommended by W3C [5] and an important language for the Semantic Web [6]. The framework comprises two major components: Model Ontology and Model Schema. The former provides meanings of specific terminology agreed upon by concerned user communities and applications; the latter contains purely generic schematic descriptions of decision models. Note that here the terms “decision model” and “model” will be used interchangeably, depending on the context. A transportation problem will serve as an example throughout since it is a well-known example often used in modelbase literature.

An overview of the Semantic Web and OWL is introduced in Section 2. Requirements for a good framework for modelbases are identified in Section 3. 4 The proposed data model: conceptual framework, 5 The proposed data model: detailed framework, 6 Operations on decision models present the essence of the proposed framework, its detailed representation, and its operations, respectively. Section 7 explains a prototype implementation, Section 8 discusses the benefits of ODDM, and Section 9 reviews related works. Finally, Section 10 contains conclusions and suggestions for future work.

Section snippets

The Semantic Web

The Semantic Web [6] is a vision for the future of the Web on which information can be processed by machines. It extends current human-readable web contents by encoding their explicit meanings in a form that can be understood by computers. As a result, machines can perform operations on web resources in an intelligent manner. The Semantic Web relies on an important technology, i.e., ontology. Ontology—a formal, explicit, shared specification of a conceptualization—provides the Semantic Web with

Requirements for good data models

A good data model or framework for the modelbase of a Web-based DSS should have the following capabilities:

  • (1)

    Compatibility with Web technologies. A Web-based DSS is obligated to employ Web-applicable technologies in most, if not all, of its components. A framework for modelbases must be Web-technology-compatible in order to yield modelbases deployable in the Web-based environment.

  • (2)

    Support of interoperability. In the Web environment, modelbases can be independently developed and maintained at

The proposed data model: conceptual framework

Generic decision problems normally contain terminologies and descriptions of the problem structure. Consider, for example, a transportation problem, the objective of which is to minimize the total cost of transporting product requirements without exceeding supplier capacities. The problem contains terms such as transportation flow, source/destination location, and shipment quantity. The problem also has description of the problem's constructs, e.g., model inputs include transportation cost of

The proposed data model: detailed framework

Due to its high expressive power, OWL Declarative Description (ODD) is employed as an underlying language to describe the framework components as well as constraints, in a formal and declarative manner.

Operations on decision models

ODD is used as a query language for the framework in addition to its use as a representation language. Therefore, not only modelbase components can be uniformly represented under the same language, operations on all components can be seamlessly performed using the same query language as well.

There exists a limited number of query languages for modelbases and existing ones inadequately address the retrieval of query results beyond a syntactic-level query [14], [28]. This framework employs an OWL

System implementation

A prototype based on the proposed framework has been implemented (Fig. 11). It is available at http://krwin.cs.ait.ac.th/mmsdss. Asp.net is employed to create Web forms and graphic user interfaces. Protégé [32] is used as an ontology editing tool to create and manage OWL elements representing Model Ontology and Model Schema. The system employs an XML rule language—XML Equivalent Transformation (XET) [2]—as the underlying language for computation, which allows direct, succinct, and efficient

Benefits of the proposed framework

ODDM can satisfy the requirements presented in Section 3 as follows:

  • (1)

    Compatibility with Web technologies-By employing ODD—a formal XML-based language deployable on the Web—as a basis, ODDM is compatible with various flavors of XML-based Web technologies such as OWL, RDF, XSLT, and Web services framework.

  • (2)

    Support of interoperability—Each component of the framework supports interoperability. First, Model Ontology enables a large degree of interoperability in terms of terminology sharing and reuse

Related works

The representation of modelbases has long been studied and the proposed framework accommodates benefits of existing research. Existing frameworks can be distinguished along two criteria: (1) the focus of content and (2) the representation techniques of decision models. A framework normally employs a combination of an appropriate focus of content and a suitable corresponding representation.

The focus of content leads to five focusing approaches. First the data-centric approach treats information

Conclusions and future works

The paper has presented a formal framework for modelbases, namely ODDM which captures key characteristics of a modern modelbase framework, facilitates the reuse and sharing of decision models, as well as handling both current and future impacts on modelbase representations. The framework yields a declarative specification delivering a user- and machine-vocabulary with clear semantics by means of ontology. This XML-based framework also permits an extraction of implicit information embedded in

Thadthong Bhrammanee is a Doctoral degree candidate in the Computer Science & Information Management Program, School of Engineering and Technology, Asian Institute of Technology, Thailand. She received a B.A. degree in Business Administration from Mahidol University, Thailand, and an M.B.A in Information Systems from the University of Toledo, USA. Her current research interests are in the area of decision support systems, and data and knowledge management.

References (51)

  • C. Anutariya et al.

    RDF Declarative Description (RDD): a language for metadata

    Journal of Digital Information

    (2001)
  • C. Anutariya et al.

    Semantic Web modeling and programming with XDD

  • D. Asuman et al.

    Enhancing ebXML registries to Make them OWL Aware

    Distributed and Parallel Databases

    (2005)
  • S. Banerjee et al.

    Model type selection in an integrated DSS environment

    Decision Support Systems

    (1993)
  • S. Bechhofer et al.

    OWL Web Ontology Language reference

  • T. Berners-Less et al.

    The SemanticWeb

    Scientific American

    (2001)
  • H. Bhargava et al.

    The World Wide Web: Opportunities for Operations Research and Management Science

    INFORMS Journal on Computing

    (1998)
  • H. Bhargava et al.

    Decision support systems and web technologies: a status report

  • H. Bhargava et al.

    Model management in electronic markets for decision technologies: a software agent approach

  • M. Binbasioglu

    Process-based constructive approach to model building

    Decision Support Systems

    (1994)
  • R. Blanning

    An entity-relationship approach to model management

    Decision Support Systems

    (1986)
  • A. Brooke et al.

    GAMS: A User's Guide (GAMS Development Corporation

    (1998)
  • J. Choobineh

    A diagramming technique for representation of linear models

    OMEGA International Journal of Management Science

    (1991)
  • J. Choobineh

    SQLMP: a data sublanguage for representation and formulation of linear mathematical models

    ORSA Journal on Computing

    (1991)
  • G. Collaud et al.

    gLPS: graphical tool for the definition and manipulation of linear problems

    European Journal of Operational Research

    (1994)
  • DAML Ontology Library,...
  • DAML Tools,...
  • O. Ezechukwu et al.

    OOF: Open Optimization Framework, Departmental Technical Report 2003/7

    (2003)
  • R. Fourer

    Database structure for mathematical programming models

    Decision Support Systems

    (1997)
  • R. Fourer et al.

    AMPL: A Mathematical Programming Language

    Management Science

    (1990)
  • GAMS model library,...
  • A. Geoffrion

    An introduction to structured modeling

    Management Science

    (1987)
  • A. Geoffrion

    Indexing in modeling language for mathematical programming

    Management Science

    (1992)
  • D. Ghosh et al.

    Model selection and sequencing in Decision Support Systems, Omega

    International Journal of Management Science

    (1991)
  • H. Greenberg

    The role of software in optimization and operations research

  • Cited by (11)

    • An assertive reasoning method for emergency response management based on knowledge elements C4.5 decision tree

      2019, Expert Systems with Applications
      Citation Excerpt :

      Most research on knowledge reasoning focuses primarily on three aspects, namely model class selection, model selection processes, and model instance determination (Ghosh & Agarwal, 1991). Related methods can be roughly divided into two camps, the analytic method and artificial intelligence method (Bhrammanee & Wuwongse, 2008). The analytic method uses numerical correlation analyses and statistics to choose (Agarwal, Tanniru, Zhang, & Agarwal, 1995).

    • A semantic service-oriented architecture for distributed model management systems

      2013, Decision Support Systems
      Citation Excerpt :

      A number of relevant design characteristics have been considered in this architecture. They are primarily driven by the issues and requirements for supporting model management during decision support in distributed settings: (1) a single model representation format [26], (2) representational independence of model structure and the detailed data [26,45], (3) representational independence of model structure and the model solution [26,45], (4) meta-modeling capability to support reasoning about models [45], (5) extensible for different modeling paradigms [26], (6) accessibility of decision support resources [21], and compatibility with the web [10]. This architecture builds on earlier work on distributed decision systems, with a particular emphasis on model management.

    • Evaluating probability of default: Intelligent agents in managing a multi-model system

      2011, Expert Systems with Applications
      Citation Excerpt :

      The combined use of intelligent agents and reasoning capable ontologies is proposed to solve these issues. Some research is done in applying the PA in the BI systems by using metadata structures (Bhrammanee & Wuwongse, 2008; Cheung & Babin, 2006) and intelligent agents (Kishore, Zhang, & Ramesh, 2006; Lavbič et al., 2007). The model presenting the combined use of IA using local KMS is offered by Perko and Bobek (2008a, 2008b), but the unified application results of the PD evaluation still were not presented.

    • Progress and applications for integrated sensing and intelligent decision in smart city

      2019, Cehui Xuebao/Acta Geodaetica et Cartographica Sinica
    • A loose-coupled mechanism of analysis model management for earthquake emergency response

      2017, Beijing Hangkong Hangtian Daxue Xuebao/Journal of Beijing University of Aeronautics and Astronautics
    View all citing articles on Scopus

    Thadthong Bhrammanee is a Doctoral degree candidate in the Computer Science & Information Management Program, School of Engineering and Technology, Asian Institute of Technology, Thailand. She received a B.A. degree in Business Administration from Mahidol University, Thailand, and an M.B.A in Information Systems from the University of Toledo, USA. Her current research interests are in the area of decision support systems, and data and knowledge management.

    Vilas Wuwongse is a Professor in the Computer Science & Information Management Program, School of Engineering and Technology, Asian Institute of Technology, Thailand. He received his B. Eng and M. Eng from the Department of Control Engineering, and D. Eng from the Department of Systems Science, Tokyo Institute of Technology, Japan. His research and teaching interests are in the areas of information modeling and representation, and Semantic Web. He published in several professional journals such as Journal of Intelligent Information Systems, IEEE Transactions on Knowledge and Data Engineering, and Computational Intelligence.

    View full text