人工知能学会論文誌
検索
OR
閲覧
検索
17 巻 , 4 号
Technical Papers / Special Issue: Intelligent Media and Its Applications in Education
選択された号の論文の18件中1~18を表示しています
    • |<
    • <
    • 1
    • >
    • >|
一般論文
論文
  • 小池 英勝, 赤間 清, 馬淵 浩司, 岡田 浩一, 繁田 良則
    原稿種別: 研究論文
    専門分野: 基礎・理論
    17 巻 (2002) 4 号 p. 354-362
    公開日: 2002/05/23
    ジャーナル フリー
    Representation and computation of negation is very important in problem solving in various application domains. The purpose of this paper is to propose a new approach to negation. While most theories for negation are based on the logic paradigm, this theory is constructed based on the equivalent transformation (ET) computation model, since the ET model provides us with “decomposability of programs,” i.e., a program in the ET model is a set of ET rules and can be synthesized by generating each ET rule independently of other ET rules.
    To represent negation in the ET model, a constraint is introduced as a pair of an object and a domain. A constraint becomes true when the object is specialized to a ground object within the domain. A negative constraint has a domain that is the complement of the meaning of the corresponding declarative description. Computation of negation in the ET paradigm is realized by equivalent transformation of declarative descriptions including negative constraints.
    For each negative constraint in a definite clause, a new declarative description is produced and transformed equivalently. When it is transformed to a set of unit clauses, the negative constraint is solved. Each unit clause returns a simple constraint to the “caller” clause. This paper proves two theorems that provide a basis for such equivalent transformation of negative constraints.
    抄録全体を表示
  • 伊藤 一之, 松野 文俊
    原稿種別: 研究論文
    専門分野: 創発システム
    17 巻 (2002) 4 号 p. 363-372
    公開日: 2002/05/23
    ジャーナル フリー
    Reinforcement learning is very effective for robot learning. Because it does not need priori knowledge and has higher capability of reactive and adaptive behaviors. In our previous works, we proposed new reinforcement learning algorithm: “Q-learning with Dynamic Structuring of Exploration Space Based on Genetic Algorithm (QDSEGA)”. It is designed for complicated systems with large action-state space like a robot with many redundant degrees of freedom. And we applied it to 50 link manipulator and effective behavior is acquired. However optimality and fault tolerance of the proposed algorithm were not considered and to demonstrate effectiveness of the proposed algorithm other applications are necessary. Acquiring of locomotion patterns by a multi-legged robot is a very interesting problem. As it has many redundant degrees of freedom, application of usual reinforcement learning is difficult and an optimal locomotion has not been acquired using previous reinforcement learning algorithm. And the redundancy of the robot is effective to the fault tolerance and various locomotion patterns can be acquired for adapting the faults of the legs. In this paper, we applied QDSEGA to acquiring of locomotion pattern by the multi-legged robot and considered the optimality and fault tolerance. Effective behavior has been obtained by using our proposed algorithm.
    抄録全体を表示
  • 鈴木 麗璽, 有田 隆也
    原稿種別: 研究論文
    専門分野: 創発システム
    17 巻 (2002) 4 号 p. 373-379
    公開日: 2002/05/23
    ジャーナル フリー
    The Baldwin effect is known as one of interactions between learning and evolution, which suggests that individual lifetime learning can influence the course of evolution without the Lamarckian mechanism. Since Hinton and Nowlan clearly demonstrated this effect by a simple evolutionary simulation, this effect has come to the attention not only of biologists but also of the computer scientists. The purpose of this paper is to clarify the effects of the spatial locality on the evolution of the phenotypic plasticity and the emergence of cooperative behavior in dynamic environments. We adopted a two-dimensional model of the evolution of strategies for Iterated Prisoner's Dilemma (IPD) as a dynamic environment, introduced phenotypic plasticity into strategies, and conducted computational experiments with various settings of two essential factors, one of which concerns the scale of interaction (which decides the neighboring members playing IPD), and another concerns the scale of reproduction (which decides the neighboring candidates occupying each grid in the next generation). In almost all experiments the necessary and sufficient amount of phenotypic plasticity was selected then the population finally established cooperation through the Baldwin effect, but the evolution of the phenotypic plasticity was heavily affected by the settings of these two parameters especially when both of them are small.
    抄録全体を表示
  • 赤嶺 有平, 遠藤 聡志, 山田 孝治
    原稿種別: 研究論文
    専門分野: AIアーキテクチャ・言語
    17 巻 (2002) 4 号 p. 380-389
    公開日: 2002/05/23
    ジャーナル フリー
    Cellular Automata (CA), which is a method for analyzing phenomena of complex systems, allows us to construct many kinds of simulators such as road traffic simulators. When we design a CA model, we need a CA simulator for verification of its result. However, at the present, we need to develop a software simulator since CA dedicated computers have not been generalized yet. We need to use a trial and error method with the simulator for design of local rules, which is the main process of making CA models. We have developed an interpreter that makes this process easily. Giving the system a local rule described in a CA dedicated language, the user can instantly confirm the result. When the user change the rules, the system reflects his changes immediately. Our system does not regulate to design the local rules belonging to the CA category because the system is a general purpose CA simulator that adopts the method by description in the special language for definitions of local rules. We can modify a rule dramatically since the system interprets a rule description language in runtime. Furthermore, we can change variable parameters using GUI interfaces. Whereby they iterate to change a rule, a conflictive conditional statement sometimes occurs in a description of local rules. This conflict is far from seeking generally. Our system detects such a conflict automatically and notifies it to users as an error.
    抄録全体を表示
  • 横尾 真, 櫻井 祐子, 松原 繁夫
    原稿種別: 研究論文
    専門分野: エージェント・分散人工知能
    17 巻 (2002) 4 号 p. 390-397
    公開日: 2002/05/23
    ジャーナル フリー
    This paper presents a new multi-unit auction protocol (IR protocol) that is robust against false-name bids. Internet auctions have become an integral part of Electronic Commerce and a promising field for applying agent and Artificial Intelligence technologies. Although the Internet provides an excellent infrastructure for executing auctions, the possibility of a new type of cheating called false-name bids has been pointed out. A false-name bid is a bid submitted under a fictitious name.
    A protocol called LDS has been developed for combinatorial auctions of multiple different items and has proven to be robust against false-name bids. Although we can modify the LDS protocol to handle multi-unit auctions, in which multiple units of an identical item are auctioned, the protocol is complicated and requires the auctioneer to carefully pre-determine the combination of bundles to obtain a high social surplus or revenue. For the auctioneer, our newly developed IR protocol is easier to use than the LDS, since the combination of bundles is automatically determined in a flexible manner according to the declared evaluation values of agents. The evaluation results show that the IR protocol can obtain a better social surplus than that obtained by the LDS protocol.
    抄録全体を表示
  • 松村 真宏, 大澤 幸生, 石塚 満
    原稿種別: 研究論文
    専門分野: 自然言語
    17 巻 (2002) 4 号 p. 398-406
    公開日: 2002/05/23
    ジャーナル フリー
    With the increasing number of electronic documents, automatic indexing from a document is an essential approach in information retrieval systems, such as search engines. This paper proposes an automatic indexing method named PAI (Priming Activation Indexing) which extracts keywords expressing assertions of a document. The basic idea is that since an author writes a document for insisting on his/her main point, impressive terms to be born in the mind of the reader could represent the asserted keywords of the document. Our approach employs a spreading activation model to extract keywords based on the activity of terms without using corpus, thesaurus, syntactic analysis, dependency relations between terms, and the other knowledge except for stop-word list. Experimental evaluations are reported by applying PAI to both papers and the archives of a mailing-list.
    抄録全体を表示
  • 古崎 晃司, 來村 徳信, 佐野 年伸, 本松 慎一郎, 石川 誠一, 溝口 理一郎
    原稿種別: 研究論文
    専門分野: 知識ベース・知識システム
    17 巻 (2002) 4 号 p. 407-419
    公開日: 2002/05/23
    ジャーナル フリー
    Ontological engineering is a successor of knowledge engineering which has been considered as a technology for building knowledge-intensive systems. Although knowledge engineering has contributed to eliciting expertise, organizing it into a computational structure, and building knowledge bases, AI researchers have noticed the necessity of a more robust and theoretically sound engineering which enables knowledge sharing/reuse and formulation of the problem solving process itself. Knowledge engineering has thus developed into “ontological engineering” where “ontology” is the key concept to investigate. Although the necessity of an ontology and ontological engineering is well-understood, there has been few success stories about ontology construction and its deployment to date. The reason for this is that the principles of ontology design is not clear enough. Therefore, a methodology of ontology design and a computer system supporting ontology design are needed. Our research goals include a methodology of ontology design, and development of an environment for building and using ontologies. This article outlines an environment for building and using ontologies “Hozo” which is under development. Hozo is designed based on a fundamental consideration of an ontological theory. And it has been extensively used in many projects to develop various ontologies. As an example, this paper presents an activity of ontology construction and its deployment in an interface system for an oil-refinery plant operation which has been done under the umbrella of Human-Media Project for five years. And we demonstrate the ability of Hozo through the ontology building/using process.
    抄録全体を表示
  • 土屋 雅稔, 黒橋 禎夫, 佐藤 理史
    原稿種別: 研究論文
    専門分野: 自然言語
    17 巻 (2002) 4 号 p. 420-430
    公開日: 2002/05/23
    ジャーナル フリー
    This paper proposes a method to discover definitoon patterns automatically from an ordinary dictionary. A definition pattern, which is frequently used to describe words and concepts in a ordinary dictionary, determines a set of similar words and can be used as a template to clarify distinctions among them. To discover these definition patterns, we convert definition sentences into tree structures, and compress them using the MDL principle. The experiment on a Japanese children dictionary is reported, showing the effectiveness of our method.
    抄録全体を表示
  • 大谷 紀子, 大和田 勇人
    原稿種別: 研究論文
    専門分野: 学習と発見
    17 巻 (2002) 4 号 p. 431-438
    公開日: 2002/05/23
    ジャーナル フリー
    This paper describes a method for optimal hypothesis search in Inductive Logic Programming(ILP). The method is based on symbiotic evolution, a variant of genetic algorithm (GA), for improving the predictive accuracy in classifying unknown examples. Progol, the representative ILP system, employs a refinement operator and finds an optimal hypothesis which subsumes the most specific hypothesis. Progol focuses on a hypothesis which has maximum explanatory power for training data. However, ILP systems should be evaluated by their explanatory powers for unknown data.
    In contrast, the proposed method produces a hypothesis using symbiotic evolution, which maintains and evolves two populations: a population of partial solutions to the problem and a population of complete solutions which are formed by grouping several partial solutions together. Symbiotic evolution can conduct balanced optimization of partial solutions and complete solutions. We postulate that the diversity of the results in GA increases the fitness to unknown data.
    We have developed an ILP system called ILP/SE, which uses symbiotic evolution for the hypothesis search task and uses the learning algorithm of Progol for the other task. ILP/SE judges the class of unknown data by majority using multiple hypothesises obtained in repeated execution. Experiments were conducted to show the performance of ILP/SE using the mutagenesis dataset. The result indicates that the ILP/SE approach outperforms the previous method using Progol for classification accuracy.
    抄録全体を表示
特集論文
論文
  • Takeuchi Yugo, Katagiri Yasuhiro
    原稿種別: 特集論文
    専門分野: ヒューマンインタフェース・知的学習支援システム
    17 巻 (2002) 4 号 p. 439-448
    公開日: 2002/05/30
    ジャーナル フリー
    Social persuasion abounds in human-human interactions. Attitudes and behaviors of people are invariably influenced by the attitudes and behaviors of other people as well as our social roles/relationships toward them. In the pedagogic scene, the relationship between teacher and learner produces one of the most typical interactions, in which the teacher makes the learner spontaneously study what he/she teaches. This study is an attempt to elucidate the nature and effectiveness of social persuasion in human-computer interaction environments. We focus on the social dynamics of multi-party interactions that involve both human-agent and inter-agent interactions. An experiment is conducted in a virtual web-instruction setting employing two types of agents: conductor agents who accompany and guide each learner throughout his/her learning sessions, and domain-expert agents who provide explanations and instructions for each stage of the instructional materials. In this experiment, subjects are assigned two experimental conditions: the authorized condition, in which an agent respectfully interacts with another agent, and the non-authorized condition, in which an agent carelessly interacts with another agent. The results indicate performance improvements in the authorized condition of inter-agent interactions. An analysis is given from the perspective of the transfer of authority from inter-agent to human-agent interactions based on social conformity. We argue for pedagogic advantages of social dynamics created by multiple animated character agents.
    抄録全体を表示
  • 松居 辰則, 石川 智剛, 岡本 敏雄
    原稿種別: 特集論文
    専門分野: 学習と発見
    17 巻 (2002) 4 号 p. 449-461
    公開日: 2002/05/30
    ジャーナル フリー
    The purpose of this study is development of a supporting system for teacher's design of lesson plan. Especially design of lesson plan which relates to the new subject "Information Study" is supported. In this study, we developed a system which generates teaching and learning activity sequences by interlinking lesson's activities corresponding to the various conditions according to the user's input. Because user's input is multiple information, there will be caused contradiction which the system should solve. This multiobjective optimization problem is resolved by Distributed Genetic Algorithms, in which some fitness functions are defined with reference models on lesson, thinking and teaching style. From results of various experiments, effectivity and validity of the proposed methods and reference models were verified; on the other hand, some future works on reference models and evaluation functions were also pointed out.
    抄録全体を表示
  • 堀口 知也, 平嶋 宗
    原稿種別: 特集論文
    専門分野: ヒューマンインタフェース・知的学習支援システム
    17 巻 (2002) 4 号 p. 462-472
    公開日: 2002/05/30
    ジャーナル フリー
    In simulation-based learning environments, 'unexpected' phenomena often work as counterexamples which promote a learner to reconsider the problem. It is important that counterexamples contain sufficient information which leads a learner to correct understanding. This paper proposes a method for creating such counterexamples. Error-Based Simulation (EBS) is used for this purpose, which simulates the erroneous motion in mechanics based on a learner's erroneous equation. Our framework is as follows: (1) to identify the cause of errors by comparing a learner's answer with the problem-solver's correct one, (2) to visualize the cause of errors by the unnatural motions in EBS. To perform (1), misconceptions are classified based on problem-solving model, and related to their appearance on a learner's answers (error-identification rules). To perform (2), objects' motions in EBS are classified and related to their suggesting misconceptions (error-visualization rules). A prototype system is implemented and evaluated through a preliminary test, to confirm the usefulness of the framework.
    抄録全体を表示
  • 村上 正行, 角所 考, 美濃 導彦
    原稿種別: 特集論文
    専門分野: ヒューマンインタフェース・知的学習支援システム
    17 巻 (2002) 4 号 p. 473-480
    公開日: 2002/05/30
    ジャーナル フリー
    In this article, we report our analysis of how the students' eye movement is affected by the content of lecture in order to utilize as standard of selection of image for distance learning and WBT. We classified content of lecture into nine parts: introduction, presentation, explanation, illustration, assertion, query, reply, question, response.We analyzed students' eye movement in the multimedia lecture "Japanese Economics", which was distance lecture between Kyoto University and UCLA.
    As the result of analysis, we get the following characteristic of eye movement of each course process in practical lecture.Introduction; students gaze at lecturer at first in order to achieve advance organizer, and next look at material.Presentation; they mainly stare at material and sometimes peer at lecturer to complement lack of understanding with information given by lecturer.Explanation; staring time is longer than other course process categories, and students stare at the object which they regard as important.Illustration; students stare at material which offers main information source.Assertion; they gaze at lecturer because of interaction between lecturer and students.Question-and-answer; generally students look at speaker but in the case of "query" about material, they change their focuses on material and lecturer fast and by turns in order to get information of lecturer and material.And our research suggests the practical guide for our choice of image information.
    抄録全体を表示
  • 山田 博文, 松田 和彦, 田口 亮, 桂田 浩一, 小林 聡, 新田 恒雄
    原稿種別: 特集論文
    専門分野: マルチメディア
    17 巻 (2002) 4 号 p. 481-489
    公開日: 2002/05/30
    ジャーナル フリー
    This paper describes a method for extracting importance of slides in a lecture review system. We introduce "index of importance" to quantitatively evaluate importance of slides. The index of importance is subjective evaluation value that is attached to each slide by lecturers. Firstly, the lecture review system extracts the index of importance of the slide by using a multi-layer neural network (MLN). In a MLN learning process, eight types of nonlinguistic informations, such as the presentation time of the slide, are used as inputs and the index of importance given by lecturers are set as outputs. Secondly, the index of importance is modified by using the other MLN which has two types of inputs; one is the index of importance and the other is similarities between the slide and adjacent slides. The similarities are calculated with key-word vectors extracted by linguistic informations in slides. The experimental results showed that the index of importance extracted by the system is highly correlated with the index attached by lecturers. As a result, the lecture review system with the proposed extraction method can properly detect key slides and helps students to learn the contents of a lecture effectively.
    抄録全体を表示
  • 中池 竜一, 三輪 和久
    原稿種別: 特集論文
    専門分野: ヒューマンインタフェース・知的学習支援システム
    17 巻 (2002) 4 号 p. 490-499
    公開日: 2002/05/30
    ジャーナル フリー
    In the present study, we show a simulated experiment environment, VPL(Virtual Psychology Laboratory), for visualizing user's exploratory experimental behavior, and present two main modules of the environment: (1) a cognitive simulator and (2) a system for automatically describing experimenter's behavior based on EBS (Exploratory Behavior Schema) proposed by the author. Users use this environment as an experimental psychologist who investigates human collaborative discovery. They experience many trials of conducting experiments using the simulated environment, and analyze by themselves their experimental processes based on the description of their behavior by EBS. It is expected that learners can notice their errors of experimental planning and refine various types of knowledge related to the experimental skills by repeating the experimental activities in this environment.
    抄録全体を表示
  • 小林 聡, 田中 敬志, 森 一将, 中川 聖一
    原稿種別: 特集論文
    専門分野: ヒューマンインタフェース・知的学習支援システム
    17 巻 (2002) 4 号 p. 500-509
    公開日: 2002/05/30
    ジャーナル フリー
    Many language learning materials have been published. In language learning, although repetition training is obviously necessary, it is difficult to maintain the learner's interest/motivation using existing learning materials, because those materials are limited in their scope and contents. In addition, we doubt whether the speech sounds used in most materials are natural in various situations. Nowadays, some TV news programs (CNN, ABC, PBS, NHK, etc.) have closed/open captions corresponding to the announcer's speech. We have developed a system that makes Computer Assisted Language Learning (CALL) materials for both English learning by Japanese and Japanese learning by foreign students from such captioned newscasts. This system computes the synchronization between captions and speech by using HMMs and a forced alignment algorithm. Materials made by the system have following functions: full/partial text caption display, repetition listening, consulting an electronic dictionary, display of the user's/announcer's sound waveform and pitch contour, and automatic construction of a dictation test. Materials have following advantages: materials present polite and natural speech, various and timely topics. Furthermore, the materials have the following possibility: automatic creation of listening/understanding tests, and storage/retrieval of the many materials. In this paper, firstly, we present the organization of the system. Then, we describe results of questionnaires on trial use of the materials. As the result, we got enough accuracy on the synchronization between captions and speech. Speaking totally, we encouraged to research this system.
    抄録全体を表示
  • 柏原 昭博, 鈴木 亮一, 長谷川 忍, 豊田 順一
    原稿種別: 特集論文
    専門分野: ヒューマンインタフェース・知的学習支援システム
    17 巻 (2002) 4 号 p. 510-520
    公開日: 2002/05/30
    ジャーナル フリー
    Web-based learning resources provide learners with hyperspace where they can navigate in a self-directed way to learn the contents included in the Web pages. The navigation involves making a sequence of the pages visited, which is called navigation path. However, learners often fail in making the navigation path due to a cognitive overload, which is caused by diverse cognitive efforts at comprehending the contents in Web pages, and monitoring the navigation process such as planning and reflection of navigation path. In the self-directed learning, in particular, it is difficult for learners to maintain the navigation monitoring. Focusing on navigation planning, this paper addresses the issue of how to facilitate learners' navigation monitoring to promote their learning in hyperspace provided by Web-based learning resources. Our approach to this issue is to provide learners with a space, in which they can see through the learning resources to plan a navigation path, apart from hyperspace. In this paper, we also demonstrate an assistant system for the navigation planning, which is composed of hyperspace map, page previewer, and path previewer. These facilities give learners an overview of Web pages and navigation path to help them make a navigation path plan without visiting Web pages in hyperspace. This paper also describes a case study with the assistant system. The results indicate that the system facilitates navigation particularly in a more complicated hyperspace.
    抄録全体を表示
  • 國近 秀信, 宇留島 稔, 平嶋 宗, 竹内 章
    原稿種別: 特集論文
    専門分野: ヒューマンインタフェース・知的学習支援システム
    17 巻 (2002) 4 号 p. 521-529
    公開日: 2002/05/30
    ジャーナル フリー
    This paper describes a definition of complexity of questions for a question and answer function in an intelligent support system for English learning, and its evaluation. To realize adaptive question and answer, the system should generates questions depending on both educational intentions and the learner's understanding state. For generating suitable questions for the learner automatically, we must investigate the factors which influence difficulty of questions, and prepare the mechanism to calculate the difficulty. The difficulty is composed of the learner dependent part and the independent part. The former is evaluated by referring to a student model. The latter is defined by enumerating factors which influence complexity of questions. We present a definition of the complexity along with learners' answering flow; understanding text sentences, understanding a question and composing an answer. Moreover, we describe experimentation comparing the complexity of questions calculated by computer according to the definition with the complexity evaluated by human.
    抄録全体を表示
    • |<
    • <
    • 1
    • >
    • >|
feedback
Top