Kazuhiko Nakayama

Generation of a pseudothesaurus for information retrieval based on cooccurrences and fuzzy set operations

A thesaurus in bibliographic information retrieval is a list of technical terms with relations among them, enabling generic retrieval of documents having different but related keywords. Since the construction of a thesaurus is resource consuming an automatic generation method of a thesaurus-like structure is needed. A set-theoretical model of an abstract thesaurus is developed which is related to an automatic generation method based on cooccurrences of terms in the set of texts. Replacement of a basis set in the model and transformation of cooccurrence frequencies into fuzzy sets enables the transition from the abstract mathematical model to an actual procedure of automatic generation. The generated structure is called a pseudothesaurus. An algorithm to generate the pseudothesaurus from a large amount of data is developed. Moreover, two examples based on a dictionary of scientific usage and on an actual bibliographic database are given.

Fuzzy Information Retrieval Based on a Fuzzy Pseudothesaurus

A fuzzy bibliographic information retrieval based on a fuzzy thesaurus or on a fuzzy pseudothesaurus is described. A fuzzy thesaurus consists of two fuzzy relations defined on a set of keywords for the bibliography. The fuzzy relations are generated based on a fuzzy set model, which describes association of a keyword to its concepts. If the set of concepts in the fuzzy set model is replaced by the set of documents, the fuzzy relations are called a pseudothesaurus, which is automatically generated by using occurrence frequencies of the keywords in the set of documents. The fuzzy retrieval uses two fuzzy relations in addition, that is, a fuzzy indexing and a fuzzy inverted file: the latter is the inverse relation of the former. They are, however, related to different algorithms for indexing and retrieval, respectively. An algorithm of ordering retrieved documents according to the values of the fuzzy thesaurus is proposed. This method of the ordering is optimal in the sense that one can obtain documents of maximum relevance in a fixed time interval. An example of the fuzzy retrieval is shown on a prototype database. This method shows one of the simplest way to realize fuzzy retrieval in practical database systems.

Similarity Measures Based on a Fuzzy Set Model and Application to Hierarchical Clustering

A fuzzy set model for generalizing similarity measures of binary characters for numerical classification is proposed. A set-theoretical model representation is given for well-known similarities of binary characters in mathematical taxonomy. Then a fuzzy extension of the set-theoretical model leads to generalizations of these similarities. Moreover an algorithm of hierarchical agglomerative clustering is developed in which similarity between a pair of clusters is calculated by referring to the model. An example based on psychological association is shown.

Directed Graph Representations of Association Structures: A Systematic Approach

Methods for analyzing structures of psychological associations in a systematic way are developed. A framework is developed which shows relations among the methods. First, a directed graph representation and a cluster analysis procedure are considered for a particular application for which the data was obtained in a controlled association test. Second, the directed graph and the cluster analysis serve as a basis for developing two methods for statistical hypothesis testing. These hypothesis-testing methods are used for comparing association structures between two populations. Finally, a framework showing the relation among these four methods is used for developing a different system of methods for a free association test. These methods are applied to data obtained from association tests. A questionnaire on the living conditions of local residents in urban areas was administered to obtain this data. The aim of the questionnaire is to provide macroscopic views for decisionmaking on improvement needed for the urban environment. Several diagrams showing the association structures are given, and the significance of this family of methods is discussed.

A Technique of Two-Stage Clustering Applied to Environmental and Civil Engineering and Related Methods of Citation Analysis.

A method of two‐stage clustering of literature based on citation frequency is developed. This method enables one to cluster thousands of articles Into a simplified dendrogram by introducing a two‐stage algorithm based on a citation space whose axes are cited articles. In the first step, preliminary clusters are generated which become initial clusters in the second step. Secondly, the initial clusters are connected into a dendrogram by means of an agglomerative technique. This method is applied to 5065 articles from 52 journals in environmental and civil engineering. The data are derived from the 1977 Science Citation Index [1]. Related methods of citation analysis such as hierarchical graph, clustering of journals, and multidimensional scaling are also applied to the same set of articles. These results are compared and structures of the literature in environmental and civil engineering are discussed.

A directed graph representation based on a statistical hypothesis testing and application to citation and association structures

A statistical hypothesis testing method is developed for generating a directed graph from a matrix of nonsymmetric interactions among elements of a system. Two threshold parameters are applied to an interaction matrix in order to generate edges of the graph. Chi-square goodness-of-fit test is considered. The generated graph is plotted on a plane as a figure with hierarchically order vertices. Two applications are considered: one is the structure of literature in computer applications and cybernetics based on citation relationships; the other is the structure of scientific notions of students based on an association test in educational psychology.

GAMMA-NET: A local computer network coupled by a high speed optical fiber ring bus — system concept and structure

Abstract The concept and structure of a local computer network are presented for a full scale high performance computer network coupled by a dual 32 Mbps optical fiber ring bus. The design goals of the system are efficient resource sharing and improved RAS. Implementation issues of the system are discussed, focussing on the optical fiber ring bus and the network operating system services. The experimental results also include the processing times for a data link connection/disconnection, the synchronous delay time for a read/write operation matching, and the data transmission speed and efficiency of both the multiplex mode and the burst mode.

A View of Studies on Bibliometrics and Related Subjects in Japan

This article surveys studies on bibliometrics and related subjects in Japan. Reviewed articles are classified according to the following categories: (a) studies on bibliometrics—including bibliometric laws, citation studies, scientific communication, and software tools for bibliometrics; and (b) application of bibliometrics—including policies for scientific research, bibliometrics and information retrieval, and databases in oriental languages. An interesting characteristic in the Japanese studies is that databases of texts in oriental languages such as Japanese and Chinese have been developed. Applications of fuzzy set theory to document retrieval using bibliometric techniques are also observed. We emphasize the models and methods used in common between bibliometrics and other fields of sciences.

Subtraction problem generating: two approaches

Abstract To prevent mislearning caused by having studuents face problems beyond mastered skills, instruction/test designers and computerized tutors need to produce problems specific to a set of conditions. In this paper procedural domain of multi-column subtraction is employed to illustrate some ideas in problem generating not concerned directly with detecting bugs. We describe programs based on two different techniques. The first program makes use of a production system model for carrying out written subtraction. The flow of randomly generated problems with some “static” properties is run through the model. The process of interpretation is traced, and the program picks only the problems which possess specified “dynamic” properties. The second program is given a block structure type and a set of the subtraction facts. The block structure type is a specification of some essential processing properties of the problem. Abstract interpretation of the subtraction procedure adds relations between variables which represent initial and actual column operands and results. Then all possible patterns with bindings to the required subtraction facts are produced. A concrete problem is generated as a random consistent instantiation of a chosen pattern with substitutions from the set of already mastered facts.

Response markup with an edit distance algorithm: a technique for providing learners with feedback on misspellings

Abstract Instructional systems that accept text responses entered by the learner must be capable of dealing with the inevitable occurrence of misspellings. A widely known procedure which finds the edit distance between two strings, and has been found effective in recognizing misspellings, can be easily extended to also annotate the response to give the learner feedback on the precise nature of the error. This paper considers the instructional value of text markup and shows how to extract markup information from the matrix normally generated in the calculation of edit distance. Included is a listing of a short Pascal program that illustrates the main concepts discussed.