[in Japanese]

View full abstract

　This article considers the using of XML at the archives in Japan. It also insists on the following: If information or records are managed electronic in an organization, the using of XML at the archives will control the using of XML at the organization; If the records management and the archival management are consecutive, the archives will be used for the knowledge management at the organization.

View full abstract

　The authors have studied a method to unify and manage multiple scale vectorized map data using hierarchical structure. The method manages vectorized map data by tree structure, easily extracts only necessary data according to a reduced scale of a map. Which should be displayed, and draws it. The features of the proposed method are that it draw vector data more quickly than a method of always drawing all data regardless of the reduced scale, and that the total amount of data is less than that of the method which independent data is prepared beforehand at each reduced scale. Moreover, the proposed method is quite effective in reducing the capacity of HD (disk storage device) required to store all the vector data consisting of a large number of hierarchies, that is, a lot of kinds of reduced scales. It is effective to navigation systems etc. which use Web communication. This paper describes the proposed method and its data structure, moreover it clarifies the feature of the proposed method by applying the proposed method to coastline data of the Japan Islands and evaluating the experimental results.

View full abstract

　A method in an innovation and its effects are described in a patent document. This "method-caused effect" is a sort of causal relationship. Therefore we tried to extract some causal relationships from patent documents in fiber engineering using phrase-pattern matching and term list. We obtained some causal relationships between compound and its property. We also attempted to structure knowledge extracted from the causal relationships.

View full abstract

　In this paper, a new method for inducing symbolic knowledge from empirical data is proposed. This method consists of four steps. Firstly, linguistic variables of fuzzy membership functions are simply defined using an analysis of histograms. Secondly, weights of the Hopfield network are calculated! Thirdly, the states of the Hopfield network are asynchronously updated until these states remain unchanged. In the fourth step, rules are extracted via our proposed algorithm. The experiments in three data sets show good Derformance.

View full abstract