Dynamic rule activation for Extended Belief Rule Bases

Alberto Calzada, Jun Liu, Hui Wang, Anil Kashyap

    Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

    8 Citations (Scopus)

    Abstract

    Incompleteness and inconsistent situations are common in most rule-based decision support systems (DSS). However, most rule inference methods do not provide procedures to specifically tackle and/or analyze them. This research presents a single approach for both incompleteness and inconsistency issues with a simple yet effective method. During the rule activation step, data incompleteness and inconsistency may be seen as paired situations, since the former appears due to lack of information while the latter can be represented as an excess of heterogeneous information activated. To effectively take advantage of this fact, this research presents a Dynamic Rule Activation (DRA) method, which searches for a balance between both incomplete and inconsistent situations to improve the overall performance of the DSS. Although DRA is designed as a flexible method, able to work with most similarity measures, in this research it is applied in the context of Extended Belief Rule-Bases (E-BRBs). The case studies illustrated in this research demonstrate how the use of DRA can improve the accuracy of E-BRB based decision support models. In this regard, the RIMER+ model and the simple weighted average of the activated rules were tested with and without using DRA as pre-processing method.

    Original languageEnglish
    Title of host publicationProceedings - International Conference on Machine Learning and Cybernetics
    PublisherIEEE Computer Society
    Pages1836-1841
    Number of pages6
    ISBN (Electronic)9781479902576
    DOIs
    Publication statusPublished (in print/issue) - 2013
    Event12th International Conference on Machine Learning and Cybernetics, ICMLC 2013 - Tianjin, China
    Duration: 14 Jul 201317 Jul 2013

    Publication series

    NameProceedings - International Conference on Machine Learning and Cybernetics
    Volume4
    ISSN (Print)2160-133X
    ISSN (Electronic)2160-1348

    Conference

    Conference12th International Conference on Machine Learning and Cybernetics, ICMLC 2013
    Country/TerritoryChina
    CityTianjin
    Period14/07/1317/07/13

    Bibliographical note

    Publisher Copyright:
    © 2013 IEEE.

    Keywords

    • belief rule-base
    • Decision making
    • decision support system
    • information incompleteness
    • spatial decision making
    • uncertainty
    • urban regeneration

    Fingerprint

    Dive into the research topics of 'Dynamic rule activation for Extended Belief Rule Bases'. Together they form a unique fingerprint.

    Cite this