人間の連想傾向を基にした属性の重み補正による概念ベースの精錬

抄録

In recent years, the realization of a robot that can conversation like humans using a natural language is required. It is essential that a robot can judge human common sense and associate the word from the other word. Concept-Base is a knowledge base that gathers such knowledge in specific forms. Concepts, the meanings of words, are defined by sets of attributes with weights as the significance of attributes. However, as the strong implications for the concept, larger weight is given for the attribute. Then increasing the weight of attributes which human is likely to associate, is considered to be able to approach the human associative. So this paper proposes a refining Concept-Base by weight correction of attributes based on human associative tendency.