Journal of the Japanese Society for Artificial Intelligence
Online ISSN : 2435-8614
Print ISSN : 2188-2266
Volume 14 , Issue 6
Showing 1-46 articles out of 46 articles from the selected issue
Print ISSN:0912-8085 until 2013
  • Hideyuki NAKAJIMA
    Type: Preface
    1999 Volume 14 Issue 6 Pages 933
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (158K)
  • Mitsuru ISHIZUKA
    Type: Cover article
    1999 Volume 14 Issue 6 Pages 934-935
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (161K)
  • Tomonari KAMBA, Yuichi KOIKE, Yoshiyuki KOSEKI
    Type: Special issue
    1999 Volume 14 Issue 6 Pages 936-942
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (828K)
  • Hideo UMEKI
    Type: Special issue
    1999 Volume 14 Issue 6 Pages 943-950
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (938K)
  • Takaaki HASEGAWA, Hisako ASANO, Motoyuki HORII
    Type: Special issue
    1999 Volume 14 Issue 6 Pages 951-958
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (1016K)
  • Fumihiro MARUYAMA
    Type: Special issue
    1999 Volume 14 Issue 6 Pages 959-966
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (1266K)
  • Koichi TAKEDA, Yuichi NAKAMURA, Naohiko URAMOTO
    Type: Special issue
    1999 Volume 14 Issue 6 Pages 967-975
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (1003K)
  • Takahira YAMAGUCHI
    Type: Cover article
    1999 Volume 14 Issue 6 Pages 976
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (105K)
  • Riichiro MIZOGUCHI
    Type: Special issue
    1999 Volume 14 Issue 6 Pages 977-988
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (1340K)
  • Kazuo FURUTA, Shinji YOSHIKAWA
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 989-1000
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    A generic model of experts' knowledge is required as a basis of good support tools for human-machine communication. The aim of this study is to construct a knowledge model of a plant opeator by behavior-speech analysis. An experiment was carried out in order to investigate substance of knowledge used in process control. Subjects' cognitive processes resulted in the observed performance were captured as logical formulae so that the knowledge used were elicited and classified. We adopted this approach with expectation that more reliable and elaborate analysis might be possible than classification by raw expert judgement. The analysis revealed that there are primarily four types of fundamental knowledge: state, goal, causality, and configuration. Subjects' cognitive processes in different experimental scenarios were characterized by classifying their utterances into the knowledge types and temporally tracing the processes. Consequently a knowledge model that consists of four spaces for describing knowledge of these four types and interrelations among them has been proposed.

    Download PDF (1328K)
  • Fumiaki ITOH, Noriko OTANI, Takaya UEDA, Yuji IKEDA
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1001-1009
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    We propose a two-way navigation system, Canopy, for the real space and the information space. Canopy expresses the real space with live remote images taken by Internet cameras. We move in the real space with controlling camera angles and with selecting objects on maps. Canopy also handles the Web information space. We move in the information space with selecting links to another page, with menus and with retrieving some pages. Canopy provides not only those navigations but also the one between two spaces. If we move in the real space or in the information space, Canopy navigates us in the other space. Because the structures of two spaces are different, these navigation functions are useful. Canopy automatically integrates information in both spaces to achieve two-way navigation. It extracts a hypothetical ontology form a table in a HTML document, then integrates it with the stable ontology of map information. Ontologies are integrated by the EBFM technique wer developed. We also propose this information integration method for Web pages and real space information such as maps and camera images.

    Download PDF (1128K)
  • Keiichi HIROTA, Yutaka SASAKI, Tsuneaki KATO
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1010-1018
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    This paper proposes ontology-driven information extraction (ODIE), a novel approach to extracting information from a text corpus without using domain-dependent information extraction (IE) rules. Instead, the ODIE approach employs ontology as a semantic guide for selecting key information form the texts. First, we discuss a relation between information extraction and ontology, then we define application ontology for the purpose of ODIE. Next, we describe a novel method for achieving ODIE that consists of a marker passing method, the biasing of ralation paths, and the extraction of template slot names and fillers. Experimental results of the extraction of four pieces of information, such as company names and product names from 250 newspaper articles show high precision and recall. These results strangly support the feasibility and high potential of ODIE. We also discuss a possibility of a semi-automatic construction of the application ontology and the effect of incompleteness in the ontology on the IE task.

    Download PDF (1047K)
  • Riichiro MIZOGUCHI, Mitsuru IKEDA, Yoshinobu KITAMURA
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1019-1032
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    Building an ontology requires clear understanding of what can be concepts with what relations to other concepts. an ontology thus focuses on "objects" themselves rather than "representation" of them. However, few research have dealt with what to represent to date. Theory of content (what to represent) is badly needed. As a basis of ontological engineering, this paper presents ontological theories of semantic links such as is-a and part-of based on the set theory. Formulation of instance-of, is-a and part-of links is done based on member-of and subset-of relations sets. Concepts of relations and roles are also investigated to come up with a few guidelines of an ontology design. Although the research is not completed yet, basic design for an ontology representation language of solid foundation has been done.

    Download PDF (1518K)
  • Keiji HIRATA, Rumi HIRAGA
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1033-1040
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    This paper discuses the roles of a musical ontology for performance rendering application. Basically, performance rendering consists of two stages: score analysis and the application of a performance model to the results of the score analysis. However, few attempts have been successful at programming these two stages to realize natural and expressive performance. This is because the characteristics of music are tacit, incomplete, and subjective, and accordingly, almost all conventional systems employed ad hoc methods with heuristics. While a few exceptional systems work well, they do so in very limited musical genres. We introduce a musical ontology to build a performance rendering system. Since the musical ontology allows us to treat music in a formal and explicit way, it will enable musical concepts to be real entities in a computer program. With the musical ontology, we can elaborate cause-and-effect relationships among the score analysis, performance model, and system output. In our paper, we will first formalize the problem of performance rendering and survey the current situation of musical ontology research. Then, we will describe the application ontology for performance rendering, which is based on a deductive object-oriented database technique.

    Download PDF (908K)
  • Masahiro HORI
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1041-1050
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    This paper presents the concept of a domain-oriented library of problem-solving methods, which consists of loosely coupled three modules for a domain ontology, problem-solving methods, and graphical user interfaces. A domain ontology provides fundamental concepts and their relations to be employed by the other two modules. Relations among concepts in a domain ontology help to articulate relationships between top-level problem-solving methods, which can not be clarified relying solely on data flow or control structure as in the conventional task-method decomposition hierarchies. A domain-oriented library of production scheduling methods, SCOOP, is then introduced, and its real-life application systems are briefly explained. Finally, empirical evaluation of the domain ontology in SCOOP is conducted, taking account of attributes and relations defined for domain concepts in the real applications.

    Download PDF (1200K)
  • Takayuki SEKIYA, Masaharu YOSHIOKA, Tetsuo TOMIYAMA
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1051-1060
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    During design, a designer uses various computational tools, such as a geometric modeling system, analysis tools, and databases. To support these design processes, a system that can integrate such computational tools is required. For this purpose, we have been developing the Knowledge Intensive Engineering Framework (KIEF) system. KIEF enables the designer to build, evaluate, and modify design models on the system by providing him/her with a wide variety of engineering knowledge. A common ontology plays an important role in integrating tools and representing engineering knowledge. This paper describes the structure of the ontology in KIEF, and how it is used in a modeling processes on KIEF.

    Download PDF (1264K)
  • Noriaki IZUMI, Akira MARUYAMA, Atsuyuki SUZUKI, Takahira YAMAGUCHI
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1061-1071
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    This paper proposes a development process of expert systems by using a library of re-usable methods which supports the processes both in specifications and in implementations for the development of expert systems. The objectives of this paper are to integrate the processes of specification and implementation into a single process, and to cut off the cost of the development of expertsystems. The methods of the library are organized into an executable specification language REPOSIT (REusable Pieces Of Specification-Implementation Templates) which combines two types of expressions: one is the relationship among input, output and reference knowledge, and the other is a prolog-based representation. A specification description of expertsystems is given as a knowledge-flow graph which is comprised of REPOSIT methods and knowledge names. Each method consisting an expertsystem's specification can be refined into a combination of fine-grained methods. In order to support the step by step operationalization of abstract models into detailed implementation descriptions, the following methods are given: a) classifying the adding patterns of control structures, b) standardizing the knowledge management and c) providing refinement policies. Finally, through the application of our proposed method to expertsystems' development, we have confirmed that REPOSIT improves the re-usability of implementation codes in the maintenance of expert systems

    Download PDF (1353K)
  • Tomohiro NAKATANI, Hiroshi G. OKUNO
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1072-1079
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    This paper presents that sound ontology plays an important role as a common vocabulary for sound representation and as a common terminology for integrating various sound stream segregation systems. Since research on computational auditory scene analysis (CASA) focuses on recognizing and understanding various kinds of sounds, sound stream segregation which extracts each sound stream from a mixture of sounds is essential for CASA. Even if sound stream segregation systems use a harmonic structure of sound as a cue of segregation, it is not easy integrate such systems because the definition of a harmonic structure differs or the precision of extracted harmonic structures differs. Therefore, sound ontology is needed as a common knowledge representation of sounds. Another problem is to interface sound stream segregation systems with applications such as automatic speech recognition systems. Since the requirement of the quality of segregated sound streams depends on applications, sound stream segregation systems must provide a flexible interface. Therefore, sound ontology is needed to fulfill the requirements imposed by them. In addition, the hierarchical structure of sound ontology provides a means of controlling top-down and bottom-up processing of sound stream segregation.

    Download PDF (1013K)
  • Takahira YAMAGUCHI, Masaki KUREMATSU, Chizuru AOKI, Rieko SEKIUCHI, Sh ...
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1080-1087
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    This paper focuses on how to construct domain ontologies, in particular, a hierarchically structured set of domain concepts without concept definitions, reusing a machine readable dictionary (MRD) and making it adjusted to specific domains. In doing so, we must deal with concept drift, which means that the senses of concepts change depending on application domains. So here are presented the following two strategies: match result analysis and trimmed result analysis. The strategies try to identify which part may stay or should be moved, analyzing spell match results between given input domain terms and a MRD and the result that trivial internal nodes have been deleted from spell match results. We have done case studies in the filed of some law. The empirical results show us that our system can support a user in constructing a domain ontology.

    Download PDF (1006K)
  • Noriaki YOSHIURA, Naoki YONEZAKI
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1088-1099
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    The formalization of human deductive reasoning is a main issue in artificial intelligence. Although classical logic (CL) is one of the most useful ways for the formalization, the implication of CL has some fallacies. For example, in Cl, A→B can be inferred form B for and arbitrary formula A. This inference is incorrect from the viewpoint of the meaning of implication which human has. In human deductive reasoning, when A⇾B is inferred, A and B should be related. Relevant logic has been studied for removal of implication fallacies in CL. The system R is a typical logical system from which fallacies of relevance and validity are removed. ER is a relevant typical logical system from which fallacies of relevance and validity are removed. ER is a relevant logical system from which these fallacies are removed and this system is not weaker than R. Especially, it is known that disjunctive syllogism holds in ER but does not hold in R. This inference rule is considered to be natural in human reasoning. In this paper, we prove that ER is properly stronger than R. This means that, for the formalization of human deductive reasoning, ER is more suitable than other relevant logical systems. The proof consists of the following steps: First, the natural deduction systems FR and FR' are introduced. FR is a natural deduction system equivalent to R. It is proved that FR' is stronger than FR. Next, we show that he normalization theorem holds in FR' and that there is a proof of ER corresponding to each normal proof of FR'. In addition, We show the fact that there is theorem of ER which can not be inferred in R. It follows that ER is properly stronger than R.

    Download PDF (1219K)
  • Kazuya TANAHASHI, Shigeki FUKUTA, Mitsuru ISHIZUKA
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1100-1107
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    For propositional-version hypothetical reasoning (or abduction), particularly for its cost-based one, some efficient methods have been developed, which achieve polynomial-time reasoning with respect to problem size to compute a near-optimal solution. However, for predicate-version hypothetical reasoning that allows rich and compact knowledge representation, it seems difficult to find such a method with polynomial-order efficiency as long as we stick to symbolic manipulation. As a result, there has been no good efficient method so far in predicate-logic domain. A naive approach for efficient predicate-version hypothetical reasoning is to transform predicate knowledge in to propositional knowledge, and ten to apply and efficient method in propositional domain. This can be accomplished by a transformation in the Herbrand universe; however, this is impractical since a vast number of propositional clauses will be produced. We tried to use a deductive database technique, namely, QSQR method, to extract a knowledge portion being related to the proof of a given goal of hypothetical reasoning, and then to transform only this portion into propositional knowledge. Nevertheless, it can not improve reasoning efficiency enough because a considerable computational time is required for the transformation into propositional knowledge by the QSQR method, and the number of the resulting propositional clauses is still large. In this paper, we propose a knowledge reformation scheme for predicate rules such that the transformation into propositional rules is efficient and the number of the resulting propositional rules is small. This knowledge reformation is based on unfolding/folding processes for predicate rules. Although its applicable range is still limited, it is shown experimentally that, by introducing this knowledge reformation, a large amount of efficiency improvement is achieved for predicate-version cost-based hypothetical reasoning.

    Download PDF (969K)
  • Naomichi SUEDA, Takahiro KAWAMURA, Naoki KASE, Tsuyoshi SHIMADA
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1108-1115
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    We developed a distributed cooperative scheduling system that generates an appropriate schedule for the whole factory by negotiation among agents which act as schedulers for each job shop. One of the most important issues in distributed cooperative problem-solving is "what request is made to which agent". When an agent assigns jobs to resources within its own shop and is faced with difficulty in allocation due to already assigned jobs, it requests other agents to shift forward or backword some obstructing job that has the least influence. In other words, when an agent falls in an impasse during problem solving, it sets up a subgoal to remove the cause of the impasse and throws it to other agents for their help. The system can decrease the number of times of backtracking by a "pre-inquiry" mechanism that asks the size of the influence to each agent in advance.

    Download PDF (834K)
  • Qing MA, Kiyotaka UCHIMOTO, Masaki MURATA, Hitoshi ISAHARA
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1116-1124
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    A multi-neuro tagger that uses variable lengths of contexts and weighted inputs with information gains for part of speech tagging has been proposed by the authors [Ma 98, 馬99]. It has been shown that the tagger has an accuracy higher than any of those obtained using the single neural networks with the fixed length of inputs, which indicates that the length of the context need not be chosen empirically; it can be selected dynamically instead. In this work, by introducing elastic inputs the tagger is slimed down into a single neural network which inherits the features that the multi-neuro tagger has. Computer experiments show that the new neuro tagger has an accuracy slightly higher than the original one instead. A series of comparative experiments for the new neuro tagger and various probability models, which include the frequency model (a base-line model), local n-gram model, and HMM, are further performed for evaluating the neuro tagger. These experiments show that the elastic neuro tagger is definitely far superior to these probability models.

    Download PDF (1031K)
  • Seiji YAMADA, Yukio OHSAWA
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1125-1133
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    This paper describes a novel navigation planning method that generates a plan (a sequence of Web pages) guiding concept understanding n the WWW. It also has the ability to generate operators during planning from Web pages using keyword extraction methods. When a user wants to understand a concept, it is useful to browse for relevant Web pages in the WWW. However, in general, this task is very hard because the user does not know where such Web pages are, and has to search for them in the vast WWW search space. Even with a search engine, this consumes the user's energy. To deal with this problem, we propose navigation planning to automatically generate a sequence of Web pages by which a user systematically understand a target concept. First, with a planning framework, we formalize the browsing task in the WWW. Action is defined as the understanding of a Web page, and an operator for a Web page consists of conditional/effect knowledge. Unfortunately it is impossible to prepare operators for a Web page. Hence we develop a method to generate an operator form Web page by extracting condition/effect terms with keyword extraction techniques. Then the whole planning procedure is describes. Finally we fully implement the navigation planning system and make experiments by comparing with methods using a search engine and link-tracing (like a Web robot). As results, we found out navigation planning is a promising approach to assist the concept understanding in the WWW.

    Download PDF (1057K)
  • Kentaro SHIMIZU, Einoshin SUZUKI
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1134-1145
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    In this paper, we describe psychological experiments for integration of unsupervised learning and supervised learning. This integrated learning has an advantage over simple unsupervised learning since it outputs a class which is explained by a useful classifier. However, little attention has been paid to this kind of learning. In the integrated learning, an overly complex class gives little information and a bad classifier. It is important to consider complexity of a class and goodnes of a classifier and select attributes in clustering. Therefore, attribute selection criteria in clustering has a crucial effect on the results. We conduct psychological experiments with a commercial date set in order to investigate human beings' process of building the criteria and discover useful knowledge. These experiments show that the criteria which consider complexity or property of a class are as good as the best criterion which requires domain knowledge. Based on these experiments, we build a domain-independent computational model for integrated learning. This model was applied to an agricultural data set, and the results were appreciated by the experts. Although the obtained class is relatively complex, the classifier has an accuracy of 89%.

    Download PDF (1373K)
  • Isao ONO, Hiroshi SATOH, Shigenobu KOBAYASHI
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1146-1155
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    This paper presents a new genetic algorithm (GA) for function optimization, considering epistasis among parameters. When a GA is applied to a function to minimize it, parents are expected to lie on some ponds or along some valleys that are promizing areas because of selection pressure as the search goes on. Especially when the function has epistasis among parameters, it has valleys that are not parallel to coordinate axes. In this case, we believe that a crossover should generate children along the valleys in order to focus the search on such promizing area from a view point of search efficiency. We employ the real number vector as a representation and propose the Unimodal Normal Distribution Crossover (UNDX) taking account of epistasis among parameters. The UNDX generates children near the line segment connecting two parents so that the children lie on the valley where the two parents are when the UNDX is applied to a function with epistasis among parameters. We demonstrate that the UNDX can efficiently optimize various functions including multi-modal ones and ones that have epistasis among parameters by applying he UNDX to some famous benchmark functions.

    Download PDF (1370K)
  • Kazuteru MIYAZAKI, Sachiyo ARAI, Shigenobu KOBAYASHI
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1156-1164
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    Most of multi-agent systems have been developed in the field of Distributed Artificial Intelligence (DAI) whose schemes are based on plenty of pre-knowledge of the agents' world or organized relationships among the agents. However, these kind of knowledge would not be always available. On the other hand, multi-agent reinforcement learning is worth considering to realize the cooperative behavior among the agents with little pre-knowledge. In multi-agent reinforcement learning systems, it is important to share a reward among all agents. Conventional work has used ad hoc sharing schemes. We focus on the Rationality Theorem [Miyazaki 94] of Profit Sharing and analyze how to share a reward among all profit sharing agents. Though reward sharing may contribute to improve policy and learning speeds, it is possible to damage system behavior. It is important to preserve the rationality condition that expected reward per an action is larger than zero. When an agent gets a direct reward R (R > 0), an indirect reward μR (μ >__- 0) are given to the other agents. We have derived the necessary and sufficient condition to preserve the rationality condition as following; μ<>M-1>/<M^W(1-(1/M)^W_o)(n-1)L>, where M is the maximum number of conflicting rules in same sensory input, L is the maximum number of conflicting rational rules, W is the maximum episode length of direct reward agents, W_o is reinforcement interval of indirect reward agents, and n is the number of agents. This theory is derived by avoiding the least desirable situation whose expected reward per and action is zero. Theorefore, if we use this theorem, we can experience several efficient aspets of reward sharing. Through neumerical examples, we confirm the effectiveness of this theorem.

    Download PDF (921K)
  • Kazuhisa MIWA
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1165-1176
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    In this paper, we discuss the relationship between the computational approach and the psychological approach for investigating human problem solving. We call the former approach Model Oriented Approach (BOA). We compare both approaches based on the framework of production system models. First, we demonstrate MOA through constructions simple models, where we try to match the rule modifications MOA through constructions of simple models, where we try to match the rule modifications and additions with certain representative aspects of the individual differences of each subject. Based on this correspondence, we can produce some simulations of typical behaviors. Second, we also demonstrate BOA through constructions of well fitting models, where initially we have to focus on the fit between the model's behavior and the human behavior. After obtaining satisfactory correspondence then we move on to analysis of rule modifications and additions needed for the fit. Last, we define MOA as a deductive way of investigation, and BOA as an inductive way of investigation, both of which provide us with strong methodologies for exploring human minds.

    Download PDF (1209K)
  • Naohiro MATSUMURA, Yukio OHSAWA, Masahiko YACHIDA
    Type: Technical paper
    1999 Volume 14 Issue 6 Pages 1177-1185
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS

    In daily lives, we frequently encounter new and specific accidents. It is about a specific matter that people want to consult a counselor. Previous memory-based systems which answer queries expressed in sentenses cannot answer specific questions about these specific things, because they are referred only to past questions from which they find similar questions to the current user's one and return the corresponding answers. In this paper, we present an Answer Assembling System, which combines paragraphs in past answers for answering newly asked questions. Empirical results are evaluated by the extent of user's satisfaction.

    Download PDF (1225K)
  • [in Japanese], [in Japanese], [in Japanese], [in Japanese]
    Type: Corner article
    1999 Volume 14 Issue 6 Pages 1186-1187
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (198K)
  • [in Japanese]
    Type: Other
    1999 Volume 14 Issue 6 Pages 1188-1189
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (382K)
  • [in Japanese]
    Type: Corner article
    1999 Volume 14 Issue 6 Pages 1190-1191
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (238K)
  • [in Japanese]
    Type: Corner article
    1999 Volume 14 Issue 6 Pages 1192
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (112K)
  • [in Japanese]
    Type: Corner article
    1999 Volume 14 Issue 6 Pages 1193
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (123K)
  • [in Japanese]
    Type: Corner article
    1999 Volume 14 Issue 6 Pages 1194
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (116K)
  • Type: Activity report
    1999 Volume 14 Issue 6 Pages 1195-1197
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (262K)
  • Type: Activity report
    1999 Volume 14 Issue 6 Pages 1198
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (134K)
  • Type: Activity report
    1999 Volume 14 Issue 6 Pages 1199-1200
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (209K)
  • Type: Activity report
    1999 Volume 14 Issue 6 Pages 1201-1207
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (537K)
  • Type: Activity report
    1999 Volume 14 Issue 6 Pages 1208
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (110K)
  • Type: Activity report
    1999 Volume 14 Issue 6 Pages b001-b012
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (981K)
  • Type: Other
    1999 Volume 14 Issue 6 Pages b015-b016
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (274K)
  • Type: Cover page
    1999 Volume 14 Issue 6 Pages c006
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (33K)
  • Type: Cover page
    1999 Volume 14 Issue 6 Pages c006_2
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (136K)
  • Type: Table of contents
    1999 Volume 14 Issue 6 Pages i006
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (118K)
  • Type: Table of contents
    1999 Volume 14 Issue 6 Pages i006_2
    Published: November 01, 1999
    Released: September 29, 2020
    MAGAZINES FREE ACCESS
    Download PDF (117K)
feedback
Top