ABSTRACT
Scientific Documents described in SGML tend to have diversified structures. In full-text database, this diversity causes difficulty to provide users with uniform view. We propose a grammatical method for processing those documents efficiently. This method uses lexical and syntactical analysis, after designating correspondence between elements in two logical structures. In this paper, we describe this method and also introduce our experiment, transforming documents into secondary, tuple information.

ABSTRACT
This paper describes a configuration method of a system which can provide full text database retrieval services. The system is easy to use and has advanced features, while utilizing the information services environment currently available on the Internet at most. It adopts a client software Mosaic and a server software HTTPD forthe WWW service as a basis, and combines an existing advanced textretrieval engine with an original gateway. It realizes a user interface through which users can access to document data with hierarchical structures flexibly and efficiently.

ABSTRACT
This paper describes results of the several searches for the experimental fulltext database, which contains 50 research articles written in Japanese. Each sentence in the articles is previously assinged the Categories that the author proposed as a framework for the functional structure analysis of research articles. The results are, the precison for the search strategies with the Categories is higher than the one for the ordinary searchstrategies based on co-existence of search terms in the articles. This paper also shows the example for the method to search and display the passages in the articles with their context using the Categories.

ABSTRACT
The university libraries are expected to increase their function and capability to meet the user needs in the electronic information age. International Christian university has undertakena project of electronic publishing of Report on Research Activity of its divisions, departments, institutes and faculty members. Through discussion of training, editing and publishing process, the paper aims to explore possibilities for a university library to carry out a new function.

ABSTRACT
This paper discusses a logical design of the union catalog database which is presently considered to migrate to a database server. The union catalog system is a large database system which contains dozens of millions of records, and processing performanceis the most cruicial problem for the union catalog database on a database server. In this paper, we present a table design and an index design, and discuss its transaction processing performance on an experimental system.

ABSTRACT
Because of the complexity of Japanese writing system using many thousand of logographic characters even for daily documents, commercially produced Japanese word processors have been relying on phonetic-to-character conversion systems from the phonetic input entry on regular (typewriter-like) keyboards. However, as word processors became quite common and are extensively used, the statistical nature of the conversion system is considered annoyingand stressful by many, on account of common appearance of homophones in the language. And, although still limited, there is some renewed interest in the ``direct" character input methods based on some unique character coding on regular keyboards. This note starts off as a sequel to the survery report of a decade ago on our study of coded input methods, and then moves on to review the efforts expended by the society on the input skill acquisitionby all, specially from the time of their youth, together with the accompanying necessity of supplying smaller and ergonomically designed keyboards. Finally, the note proposes more intelligent approaches to the phonetic-to-character conversion algorithms, beyond the presently commercially supplied word processor software, and also to the standardization of romanization for phonetic input methods. As an Appendix, a short summary and assessment is included of the recenttrial in a State of Minnesota district court, which is of a law suit for compensation against occupational injury purportedly caused by keyboard operations.

ABSTRACT
This paper describes a dynamic sample allocation procedure for efficient solution space search where each search point is evaluated only through computer simulations¡¥First, statistical models for two processes of generating and testing solution candidates are discussed¡¥Based on the models, a decision equationfor dynamic sample allocation is derived¡¥Then, results of preliminary experiments using normally distributed random variables are shown¡¥Finally, the proposed allocation method is combined with general-purpose communication network simulator to show the applicability of the method to parameter optimization problems in large and complex systems¡¥

ABSTRACT
In order to provide the quality of services in high-speed, virtualcircuit packet switching networks, it is important to choose an appropriate service discipline, which determines the transmission priority of packets at switching nodes. We first define the purpose and some basic features of service disciplines, then give comparison of a couple of proposed disciplineswith respects of their work-conserving feather. Moreover, we use simulation resultsto show the differencesof packet queueing delay distribution when different service disciplines are applied.

ABSTRACT
Model-based vision is expected to be one of the most powerful methods for computer vision, which can especially be suited to extract much complicated objects from images. This method uses "models" expressing geometric composition of objects and matches them to target images to realize image understanding. Geometric information is needed to be handled to actualize this method, thusgeometric reasoning plays an major role in model-based vision. This study presents the state of the art of geometric information handling in model-based vision, some recent studies concerning geometric reasoning, and an consideration on the role of geometricreasoning in model-based vision.

ABSTRACT
In human interface systems of next generation, a mechanism which can manipulate users' diverse and detail requirements is desired for flexible human-machine or human-human communications. For thispurpose, we consider that it is important to be provided a mechanism, which allows a user to instruct one's intention explicitly, flexibly and graphically. To demonstrate the importance of this view on human interface, this paper presents a drawing editor prototype system of conceptual figures, DOGS, focused on relations among objects in a figure. In the system, a user can precisely provide his intention about the relations to the system by explicitly giving graphical indications. Given relations are preserved as constraints to offer flexible modifications of drawings. DOGS can deal with two kinds of relations, primitive and macro ones. The primitive relation represents a geometric relation among objects, and the macro relation is for a conceptual relation. In our embodiment, seven kinds of primitive relations are available. The macro relation is defined by the user as a combination of several primitive relations. The macro relations express highly structured relations, i.e., relations much closer to the user's concepts. DOGS realizes a proper drawing process for conceptual figures and a flexible visual user interface for the communication from user to computer.

ABSTRACT
A new method of understanding rule generation for drawing image understanding systems using man-machine cooperations is presented.We have proposed a drawing understanding system using state transition models to realize multi-purpose drawing understanding systems. This system performs understanding process complying withunderstanding rules which have declarative form to be obtained easily by user. However, it is known to be very hard to construct rules which requires fine tuning, e.g. threshold values. This point is thought to be problematic also for other drawing understanding systems. So we make extensions to state transition type drawing understanding systems which enable to generate such rules automatically from interactions with user. This system can generate rules, which can be used by actual drawing understandingsystems, from given plain interactions using graphical user interface (GUI). It uses learning algorithm of inductive inferenceto avoid redundant interactions and earn efficient rule generation. This paper reveals that this system can easily generate rules which have required difficult fine tuning up to nowfrom several times of plain interactions.

ABSTRACT
The Japanese higher education sector's statistics, especially the expenditure data on education and R&D are in complicated relations. As the expenditure on R&D in the Management and Coordination Agency "Survey of Research and Development" contains much expenditure on education and so on, it is overestimated than the real one. The data should be rectified appropriately in order toprepare the fundamental data for science policy planning and to compare the data internationally.

ABSTRACT
In the United States, the issue of fairness or bias in educationaland psychological testing has received considerable attention since 1960s. During this period, several procedures have been developed to detect differential item functioning(DIF). In a DIF analysis, each item is examined individually to determine if it favors one demographic population over another. Focal group and reference group are compared. This study reviews various statisticalmethods of identifying items with potential DIF and procedures for DIF estimating and statistical hypotheses testing regarding DIF. Recent studies and future issues in the United States are presented.The possibility of its further applications are also discussed.

ABSTRACT
Recently the importance of the study of scientific terms have beenrecognized, and it is sometimes claimed that the study of terms should be independent of linguistics and should constitute an independent discipline called 'terminology'. Despite the growing interest in the study of terms, there have been few works that examined the essential nature of terms. Considering this situation, the author tries to review the nature and status of terms and how the study of scientific terms should be understood, referring to the general situation concerning the study of words in linguistics as a guide for examination.

ABSTRACT
We developed a medical term thesaurus for "Clinical case reports database", which is a full text database of clinical case reports.This thesaurus consists of free terms appeard in the clinical reports, and was named "free-terms-thesaurus" temporarily. We can retrieve the database more exhaustively using free terms.

ABSTRACT
Medical processes can be regurded to be consists of a number of tasks which is paralelly executed. Data acquisdition, evaluation of the data, decision making, or execution of medical treatments are the example of these tasks. Executing these tasks, variety kinds of knowledge is required. In this report, the authors have identified some kinds of medical knowledges described in a typicalmedical textbook in Japan, and identified basic data structure of the knowledge aboutrelations between diseases and signs/symptoms. The authors also discuss about necessery conversion of knowledge representation to make the knowledge machine executable.

ABSTRACT
Mankind is endowed with versatility, together with ensuing individual differences in various capabilities, in particular, sensory capacities. Congenital blindness and deafness are extreme cases of such human variability. Besides those, certain of dyslexia may be due to some disorders of the sensory nervous system. Even the left-handedness may arise from mild disorders of sensory motorsystems. The well developed language skill is one of the major factors which distinguish mankind from other animals. The most common language is the oral one together with its rendition to writing in letters, or in Braille for the blind. In addition, sign languages are somewhat unusual but also well-recognized languages by now. In this note, we shall attempt, mainly in the framework of brain science and psychophysics, a survey for nonspecialists, on the distinctiveness and peculiarity of language processing functions of the cerebrum of those sensorily dis-advantaged, especially the nonhearing, which arise in connection with the usage of special languages to satisfy theirspecific needs. Furthermore, we shall take up, from the same viewpoint, a problem normal people encounter with when they attempt to learn a sign language as their second language, and examine an approach to lessen the difficulty, by taking advantage of the way brains function when they work for the acquisition of such a new skill.

ABSTRACT
Analysis of the linguistic data recorded in 1931, at Sikuka (or Polonajsk) in the Karafuto (or Sakhalin) Island, the southern territory of which was under the control of the Japan Empire. The data includes 190 colums of five languages of "minority peoples" there, except for Ainue. In this paper, among the linguistic data recorded in katakana, one of the representatives of Japanese writing systems, the data of Yakut language is analysed. The wordsof the data are identified with correspondent ones from modern standard Yakut language. On the process of the analysis, some information on the background of lives of Siberian natives at the period is referred to and the fact that the data in some respects well traced the changes of linguistic surroundings is argued. Fromthe analysis, what dialect of Yakut language was recorded in the data is suggested.

[without abstract]