人工知能学会論文誌
Online ISSN : 1346-8030
Print ISSN : 1346-0714
検索
OR
閲覧
検索
21 巻 , 4 号
選択された号の論文の11件中1~11を表示しています
    • |<
    • <
    • 1
    • >
    • >|
論文
  • 田中 貴紘, 片上 大輔, 新田 克己
    21 巻 (2006) 4 号 p. 319-329
    公開日: 2006/04/04
    ジャーナル フリー
    This paper describes an agent that shows advices for supporting mediator on our online mediation support system. The purpose of the advice is an education of mediator, and the agent presents it instead of the teacher. In this research, at first, we defined a mediation model that is an argumentation model between 3 people. Then, we defined some advice models based on the mediation model. The advice models create advice elements. The adviser agent monitors the mediation, gathers advice elements referring to advice models, and creates advice from elements according to the mediation scene. As a result, it becomes possible that is advising instead of the teacher according to the situation, education purpose and learner's level. We inspected the effectiveness of the advice by the experiment of moot mediation.
    抄録全体を表示
  • 鈴木 大介, 内海 彰
    21 巻 (2006) 4 号 p. 330-339
    公開日: 2006/04/11
    ジャーナル フリー
    In this paper we propose an extraction-based method for automatic summarization. The proposed method consists of two processes: important segment extraction and sentence compaction. The process of important segment extraction classifies each segment in a document as important or not by Support Vector Machines (SVMs). The process of sentence compaction then determines grammatically appropriate portions of a sentence for a summary according to its dependency structure and the classification result by SVMs. To test the performance of our method, we conducted an evaluation experiment using the Text Summarization Challenge (TSC-1) corpus of human-prepared summaries. The result was that our method achieved better performance than a segment-extraction-only method and the Lead method, especially for sentences only a part of which was included in human summaries. Further analysis of the experimental results suggests that a hybrid method that integrates sentence extraction with segment extraction may generate better summaries.
    抄録全体を表示
  • 下川 哲矢, 参沢 匡将, 渡邊 恭子
    21 巻 (2006) 4 号 p. 340-349
    公開日: 2006/04/26
    ジャーナル フリー
    This paper addresses how communication processes among investors affect stock prices formation, especially emerging predictability of stock prices, in financial markets. An agent based model, called the word of mouth model, is introduced for analyzing the problem. This model provides a simple, but sufficiently versatile, description of informational diffusion process and is successful in making lucidly explanation for the predictability of small sized stocks, which is a stylized fact in financial markets but difficult to resolve by traditional models. Our model also provides a rigorous examination of the under reaction hypothesis to informational shocks.
    抄録全体を表示
  • 原田 健, 佐久間 淳, 池田 心, 小野 功, 小林 重信
    21 巻 (2006) 4 号 p. 350-360
    公開日: 2006/04/26
    ジャーナル フリー
    Many real-world problems entail multiple conflicting objectives, which makes multiobjective optimization an important subject. Much attention has been paid to Genetic Algorithm (GA) as a potent multiobjective optimization method, and the effectiveness of its hybridization with local search (LS) has recently been reported in the literature. However, there have been a relatively small number of studies on LS methods for multiobjective function optimization. Although each of the existing LS methods has some strong points, they have respective drawbacks such as high computational cost and inefficiency of improving objective functions. Hence, a more effective and efficient LS method is being sought, which can be used to enhance the performance of the hybridization.
    Pareto descent directions are defined in this paper as descent directions to which no other descent directions are superior in improving all objective functions. Moving solutions in such directions is expected to maximally improve all objective functions simultaneously. This paper proposes a new LS method, Pareto Descent Method (PDM), which finds Pareto descent directions and moves solutions in such directions. In the case part or all of them are infeasible, it finds feasible Pareto descent directions or descent directions as necessary and moves solutions in these directions. PDM finds these directions by solving linear programming problems. Thus, it is computationally inexpensive. Experiments have shown that PDM is superior to existing methods.
    抄録全体を表示
  • 小島 一晃, 三輪 和久
    21 巻 (2006) 4 号 p. 361-370
    公開日: 2006/04/28
    ジャーナル フリー
    In mathematical learning, it is important to give learners a number of problems that have various features in both of surface problem situations and deep mathematical structures. In this study, we implement a system that generates various word problems by using problem-generation episodes. Each problem-generation episode is knowledge comprising a base example problem and a new analogical instance, which is regarded as a past case where the analogical instance was generated from the example problem. Our system can generate various problems by applying the problem-generation episodes to initial problems stored in the system. In this paper, we describe our approach to generate mathematical word-problems, and perform experimental evaluations to verify whether or not our system can expand the variety of problems. The results of the experiments indicated that our system can appropriately expand the variety. We also found that it needs interactions with a knowledgeable user.
    抄録全体を表示
  • 中村 克彦, 保科 明美
    21 巻 (2006) 4 号 p. 371-379
    公開日: 2006/05/09
    ジャーナル フリー
    This paper discusses recent improvements and extensions in Synapse system for inductive inference of context free grammars (CFGs) from sample strings. Synapse uses incremental learning, rule generation based on bottom-up parsing, and the search for rule sets. The form of production rules in the previous system is extended from Revised Chomsky Normal Form A→βγ to Extended Chomsky Normal Form, which also includes AB, where each of β and γ is either a terminal or nonterminal symbol. From the result of bottom-up parsing, a rule generation mechanism synthesizes minimum production rules required for parsing positive samples. Instead of inductive CYK algorithm in the previous version of Synapse, the improved version uses a novel rule generation method, called ``bridging,'' which bridges the lacked part of the derivation tree for the positive string. The improved version also employs a novel search strategy, called serial search in addition to minimum rule set search. The synthesis of grammars by the serial search is faster than the minimum set search in most cases. On the other hand, the size of the generated CFGs is generally larger than that by the minimum set search, and the system can find no appropriate grammar for some CFL by the serial search. The paper shows experimental results of incremental learning of several fundamental CFGs and compares the methods of rule generation and search strategies.
    抄録全体を表示
  • 小林 一樹, 山田 誠二
    21 巻 (2006) 4 号 p. 380-387
    公開日: 2006/05/16
    ジャーナル フリー
    In this paper, we focus on a problem in designing for home robots. There are some design methods for conventional artifacts such as home electric appliances. However, not all methods can be applied effectively for robots. Especially, we deal with the situation in which a robot asks a user to help it by expressing its internal state. We then propose a novel design method ``motion overlapping (MO)'' by which a robot can perform human-like behavior to express its internal state. We consider that human-like behavior of a robot causes a user to understand its internal state ``mind'' easily. A small sweeping robot which performs ``back-and-forth'' motion is designed based on MO. In experiments, we compare the expressing by MO with sounding by a buzzer and lighting by a LED as conventional nonverbal expressing. We investigate effects on users' action of cooperating with the sweeping robot. We find that the expressing by MO causes most of the users to help the robot. The differences in those expressing methods are statistically significant. The results show that our proposed method is effective as one of design method for home robots.
    抄録全体を表示
  • 久保 長徳, 森下 卓哉, 下羅 弘樹, 河原林 友美, 小高 知宏, 小倉 久和, Ian FRANK, 田中 久美子, 田所 諭, 松原 ...
    21 巻 (2006) 4 号 p. 388-397
    公開日: 2006/05/23
    ジャーナル フリー
    We designed the system, Rescue-MIKE, and implemented to the RoboCup Rescue Simulation System. The Rescue MIKE system aims to communicate with large numbers of relief workers and controllers working in a rescue domain. The system has an ability to report the situation of the domain like a newscaster. Our system consists of multiple agents. The system collects information from a simulated disaster scenario, and then produces a dialogue that fits to the actions of the agents in the domain. We described the design policy, implementation of our system, and Rescue-MIKE's facility. Rescue-MIKE initial version can provide aural information that helps to explain the unfolding scenarios to onlookers. We discussed the future applications of our system, including knowledge elicitation about disaster relief control methods, automated relief support systems, and public education about the dangers of large natural disasters such as earthquakes, floods or volcanic eruptions.
    抄録全体を表示
  • 岡部 正幸, 山田 誠二
    21 巻 (2006) 4 号 p. 398-405
    公開日: 2006/05/23
    ジャーナル フリー
    Query expansion is a technique of information retrieval to select new query terms which improve search performance. Although good terms can be extracted from documents whose relevancy has already been known, it is difficult to get enough such feedback from users in practical situations. In this paper we propose a query expansion method which performs well even if a user only notifies relevancy of documents until just a relevant one is found. In order to tackle this specific condition, we introduce two refinements to a well-known query expansion method. One is the application of transductive learning to increase the amount of latent relevant documents. The other is the introduction of a modified parameter estimation method which laps the predictions of multiple learning trials in order to differentiate the importance of candidate terms for expansion. Experimental results show that our method outperforms traditional methods when an initial search fails.
    抄録全体を表示
  • 長谷川 忍, 柏原 昭博
    21 巻 (2006) 4 号 p. 406-416
    公開日: 2006/05/23
    ジャーナル フリー
    The main topic addressed in this paper is how to help learners navigate in exploring hyperspace provided by existing web-based learning resources in which they can navigate Web pages in a self-directed way to learn the domain concepts/knowledge. Such self-directed navigation involves constructing knowledge from the contents embedded in the navigated pages, along what is called the navigation path, which has been demonstrated to enhance learning. Creation of a useful navigation path influences the knowledge construction process and plays an important role in self-directed learning in the hyperspace. On the other hand, learners often fail at creating a navigation path due to cognitive overload, which is caused by diverse cognitive efforts what may be viewed as meta-cognitive activities. Such meta-cognitive activities hold the key to success in self-directed learning. Our approach to this issue is to analyze the navigation planning tasks in order to design facilities that can more readily facilitate learners' planning activities. In this paper, we provide the learners with a navigation planning environment called Advanced Planning Assistant, which helps them plan a navigation path in an adaptive way before learning the hyperspace. This planning environment calls the learners' attention to establishing the navigation path prior to and separately from learning the hyperspace. We also report preliminary case study to evaluate the usefulness of the adaptive approach proposed. From the results of the case study, we have made sure that they are useful.
    抄録全体を表示
  • 服部 聖彦, 高玉 圭樹, 村田 智, 古谷 寛, 上野 浩史, 稲場 典康, 小田 光茂
    21 巻 (2006) 4 号 p. 417-427
    公開日: 2006/06/02
    ジャーナル フリー
    This paper proposes a new method that can diagnose nodes which consist of a large scale structure system including intermitted fault and analyzes its capability through simulations for comparisons between our method and Adaptive DSD, one of on-line distributed diagnosis methods. Our method based on ideas of a disconnecter and token node. The disconnecter is a function to cope with an intermitted fault, while the token node can collect and exchange fault node informations from other token nodes. Our simulation results have confirmed that an autonomous distributed diagnosis method with token node works well in many cases, however, in some cases where the length between normal nodes is more than two links, our method cannot diagnoses correctly. In addition to this result, the following advantages of our method are founded in comparison with the Adaptive DSD: (1) our method does not need global connections among nodes but need only two link connections; (2) our method can cope with intermitted fault; (3) our method require less information exchange than Adaptive DSD; and (4) our method can get a correct fault information faster than Adaptive DSD.
    抄録全体を表示
    • |<
    • <
    • 1
    • >
    • >|
feedback
Top