A Trial of Logic Programing in the Japanese Language

  • Indo Kenryo
    THE DEPARTMENT OF ECONOMICS KANTO GAKUEN UNIVERSITY

Bibliographic Information

Other Title
  • 日本語による論理プログラミングの試み
  • ニホンゴ ニ ヨル ロンリ プログラミング ノ ココロミ

Search this article

Abstract

Prolog is a logic programing language that constructs computer programs as Horn-clauses and applies a type of refutation algorithm that incorporates a backtrack mechanism. This study presents a series of frugal and practical applications of Prolog in natural language computer programing, mainly written in Japanese. Two main approaches were used. The first approach proposes Japanese natural language logic programing based on Prolog predicates. By splitting words with particles, queries in Japanese words will be unified to the head of rule that may have system or user defined predicates, including calling a shell command as a procedure attached in that body. In particular, particle wa is utilized for both knowledge reference and knowledge assertion. For the second approach, an automated extraction of clause programs from a well-written Japanese document titled "Calculation of inheritance tax" is demonstrated. Furthermore, the first approach is modified by incorporating negation, syllogism, and several semantic aspects. Known/unknown, old/new information, and exhaustive/contrasting are incorporated to the programs for the forms of sentences X wa Y and X ga Y. This study argued that exhaustive ga and contrasting wa can be explained formally in terms of the stable conditions for a game theoretical network model that consists of alternative relations of word pairs. Furthermore, contrasting and exhaustive cognitive functions are discussed in a similar manner in resolution of rule clauses and they relate to circumscription in non-monotonic reasoning. Additionally, as related fields, agency relationships are discussed with regard to the cognitive-social context.

Journal

Details 詳細情報について

Report a problem

Back to top