• DocumentCode
    3731378
  • Title

    Non-clausal Multi-ary alpha-Generalized Resolution Principle for a Lattice-Valued First-Order Logic

  • Author

    Yang Xu;Jun Liu;Xingxing He;Xiaomei Zhong;Shuwei Chen

  • Author_Institution
    Sch. of Math., Southwest Jiaotong Univ., Chengdu, China
  • fYear
    2015
  • Firstpage
    1
  • Lastpage
    7
  • Abstract
    The present paper focuses on a resolution-based automated reasoning theory in a lattice-valued logic system with truth-values defined in a lattice-valued algebraic structure - lattice implication algebras (LIA) in order to handle automated deduction under an uncertain environment. Particularly, as a continuation and extension of the established work on binary resolution at a certain truth-value level α (called α-resolution), a non-clausal multi-ary α-generalized resolution principle and deduction are introduced in this paper for a lattice-valued first-order logic LF(X) based on LIA, which is essentially non-clausal generalized resolution avoiding the reduction to normal clausal form. Non-clausal multi-ary α-generalized resolution deduction in LF(X) is then proved to be sound and complete. The present work is expected to provide a theoretical foundation of more efficient resolution based automated reasoning in lattice-valued logic.
  • Keywords
    "Cognition","Indexes","Algebra","Lattices","Finite element analysis","Intelligent systems"
  • Publisher
    ieee
  • Conference_Titel
    Intelligent Systems and Knowledge Engineering (ISKE), 2015 10th International Conference on
  • Type

    conf

  • DOI
    10.1109/ISKE.2015.88
  • Filename
    7383015