人工知能学会論文誌
Online ISSN : 1346-8030
Print ISSN : 1346-0714
検索
OR
閲覧
検索
26 巻 , 1 号
選択された号の論文の30件中1~30を表示しています
    • |<
    • <
    • 1
    • >
    • >|
特集論文 「エージェント」
原著論文
  • 中川 博之, 大須賀 昭彦, 本位田 真一
    26 巻 (2011) 1 号 p. 1-12
    公開日: 2011/01/06
    ジャーナル フリー
    The complexity of current software systems requires the ability to adapt at run-time, and the development of self-adaptive systems is one of the recent challenges for realizing dynamic adaptation. In this paper, we focus on the adaptation based on the components connection as a fundamental adaptation, and present our implementation framework for constructing self-adaptive systems on the basis of an agent platform. We reinforce the agent platform by adding some application program interface for behavior cooperation and its autonomic activation, and connect the behaviors with components in self-adaptive systems. We also introduce an implementation guideline: a way to identify the responsibilities for control loops and implementation patterns for these responsibilities. We also demonstrate the effectiveness of our framework and guideline through the results from our implementation experiments and show how they can be used to construct self-adaptive systems by using agent platforms.
    抄録全体を表示
  • 新出 尚之, 高田 司郎, 藤田 恵
    26 巻 (2011) 1 号 p. 13-24
    公開日: 2011/01/06
    ジャーナル フリー
    In multi-agent environments, to model cooperations among autonomous agents, many notions such as mutual beliefs and joint intentions, recognition of possibilities to achieve a goal with cooperation, and team formations, should be formally represented. In the traditional BDI logics, it is hard to treat them uniformly. We show the way to treat them uniformly using the fixed-point operator of the extended BDI logic \ omatoes. We also give some examples to apply it to the proof of some behaviors of multi-agent systems.
    抄録全体を表示
  • 高橋 謙輔, 阿部 洋丈, 廣津 登志夫, 菅原 俊治
    26 巻 (2011) 1 号 p. 25-33
    公開日: 2011/01/06
    ジャーナル フリー
    In this paper, we propose a method for the dynamic migration of the IP routing points using binary particle swarm optimization (PSO) in distributed Virtual LAN environments. Virtual LAN (VLAN) is a virtualization technique of datalink layer and can construct arbitrary logical networks on top of a physical network. However, VLAN often causes much redundant traffic due to mismatch between the topology of the logical network and that of the underlying physical network. We will show that the proposed method can adaptively select the routing points dynamically according to the observed traffic patterns and thus reduce the redundant traffic. Finally, we will evaluate the proposed method using the simulation environment.
    抄録全体を表示
  • 塚本 鋭, 白山 晋
    26 巻 (2011) 1 号 p. 34-41
    公開日: 2011/01/06
    ジャーナル フリー
    Several factors that promote cooperation on scale-free networks have been studied, and the mechanisms of the evolution of cooperation are revealed by numerical simulations and theoretical works. However, influence of the power-law exponent on the evolution of cooperation has not been manifested. In our previous study, it is found by numerical simulations that there exists an optimal value of the power law exponent which promotes cooperation. In this paper, we consider the reason why there exists the optimal value, using a theoretical analysis.
    抄録全体を表示
  • 藤井 秀樹, 吉村 忍, 高野 悠哉
    26 巻 (2011) 1 号 p. 42-49
    公開日: 2011/01/06
    ジャーナル フリー
    Road traffic is a key portion of infrastructure to support mobility and transportation of human beings and goods. At the same time, it includes various kinds of risks. One of the most critical ones is a traffic accident. To evaluate traffic accidents quantitatively, we have newly developed a cognitive error model and implemented it into a multi-agent based traffic simulator. In the traffic simulator, each component creating traffic phenomena is modeled as an agent, and interaction among numerous agents simulates nonlinear behaviors of urban traffics. An actual traffic accident often occurs when a car driver overlooks something to watch, such as other cars, pedestrians, traffic signals, or obstacles. In the cognitive error model we developed, a driver agent has its own field of view and a gazing point, and cannot recognize objects off the gazing point. Through various types of test simulations using the developed simulator, we demonstrate that the simulator with the cognitive error model is a powerful tool to quantitatively evaluate traffic accidents and to discover such a dangerous situation that accidents frequently occur.
    抄録全体を表示
  • Indika H. Katugampala, Hirofumi Yamaki, Yukiko Yamaguchi
    26 巻 (2011) 1 号 p. 50-58
    公開日: 2011/01/06
    ジャーナル フリー
    Automated Trust Negotiation (ATN) has been proposed as a mechanism to establish mutual trust among strangers. While existing fundamental protocols and strategies are shown, this paper focuses on Parsimonious strategy. The most straightforward implementation of Parsimonious strategy has a very high memory consumption which may be problematic when it is used in real world environments. This paper proposes an implementation which keeps all requests in Disjunctive Normal Form (DNF) and further reduces its memory consumption by exploiting the history of the negotiation, while keeping the completeness of the strategy intact. In addition to that, proposed method provides a criterion to detect negotiation failures. Results obtained by means of simulations showed that the method proposed is effective in achieving its goals, without increasing the overall computational overhead. Theoretical analysis of the proposed method is also presented.
    抄録全体を表示
  • 平山 勝敏, 松井 俊浩, 横尾 真
    26 巻 (2011) 1 号 p. 59-67
    公開日: 2011/01/06
    ジャーナル フリー
    Distributed Lagrangian Relaxation Protocol (DisLRP) has been proposed to solve a distributed combinatorial maximization problem called the Generalized Mutual Assignment Problem (GMAP). In DisLRP, when updating Lagrange multipliers (prices) of goods, the agents basically control their step length, which determines the degree of update, by a static rule. A merit of this updating rule is that since it is static, it is easy to implement even without a central control. Furthermore, if we choose this static rule appropriately, we have observed empirically that DisLRP converges to a state providing a good upper bound. However, it must be difficult to devise such a good static rule for updating step length since it naturally depends on problem instances to be solved. On the other hand, in a centralized context, the Lagrangian relaxation approach has conventionally computed step length by exploiting the least upper bound obtained during the search and a lower bound obtained through preprocessing. In this paper, we achieve this approach in a distributed environment where no central control exists and name the resultant protocol Adaptive DisLRP (ADisLRP). The key ideas of this new protocol are to 1) compute global information with a spanning tree, 2) update step length simultaneously with a synchronization protocol, and 3) estimate lower bounds during the search. We also show the robustness of ADisLRP through experiments where we compared ADisLRP with the previous protocols on the critically hard benchmark instances.
    抄録全体を表示
  • 山下 泰央, 高橋 大志, 寺野 隆雄
    26 巻 (2011) 1 号 p. 68-75
    公開日: 2011/01/06
    ジャーナル フリー
    While the importance of financial education is recognized in recent years, the technique for deepening an understanding to pension investment management is needed. In this research, we analyze learning method of the pension investment management in consideration of liability using the business game technique. As a result of analysis, interesting phenomena -- the participant understood the learning method of the pension investment management in consideration of liability -- were seen. This shows the effectiveness of the business game technique to learning the pension investment management.
    抄録全体を表示
  • 片柳 亮太, 菅原 俊治
    26 巻 (2011) 1 号 p. 76-85
    公開日: 2011/01/06
    ジャーナル フリー
    We propose an effective method of dynamic reorganization using reinforcement learning for the team formation in multi-agent systems (MAS). A task in MAS usually consists of a number of subtasks that require their own resources, and it has to be processed in the appropriate team whose agents have the sufficient resources. The resources required for tasks are often unknown \ extit{a priori} and it is also unknown whether their organization is appropriate to form teams for the given tasks or not. Therefore, their organization should be adopted according to the environment where agents are deployed. In this paper, we investigated how the structures of network and the number of tasks affect team formations of the agents. We will show that the utility and the success of the team formation is deeply affected by depth of the tree structure and number of tasks.
    抄録全体を表示
  • 東藤 大樹, 岩崎 敦, 横尾 真
    26 巻 (2011) 1 号 p. 86-96
    公開日: 2011/01/06
    ジャーナル フリー
    This paper proposes a novel Peer-to-Peer Information Retrieval (P2PIR) method using user feedback and query-destination-learning. The method uses positive feedback information effectively for getting documents relevant to a query by giving higher score to them. The method also utilize negative feedback information actively so that other agents can filter it out with itself. Using query-destination-learning, the method can not only accumulate relevant information from all the member agents in a community, but also reduce communication loads by caching queries and their sender-responder agent addresses in the community. Experiments were carried out on multiple communities constructed with multi-agent framework Kodama. The experimental results illustrated that the proposed method effectively increased retrieval accuracy.
    抄録全体を表示
  • 小林 寛武, 峯 恒憲
    26 巻 (2011) 1 号 p. 97-106
    公開日: 2011/01/06
    ジャーナル フリー
    This paper proposes a novel Peer-to-Peer Information Retrieval (P2PIR) method using user feedback and query-destination-learning. The method uses positive feedback information effectively for getting documents relevant to a query by giving higher score to them. The method also utilizes negative feedback information actively so that other agents can filter it out with itself. Using query-destination-learning, the method can not only accumulate relevant information from all the member agents in a community, but also reduce communication loads by caching queries and their sender-responder agent addresses in the community. Experiments were carried out on both single and multiple communities constructed with multi-agent framework Kodama. The experimental results illustrated that the proposed method effectively increased retrieval accuracy.
    抄録全体を表示
  • 西村 一彦, 中川 博之, 田原 康之, 大須賀 昭彦
    26 巻 (2011) 1 号 p. 107-115
    公開日: 2011/01/06
    ジャーナル フリー
    Autonomic systems aim to reduce the configuration, operational, and maintenance tasks of large distributed applications. In order to implement autonomic systems, several approaches, such as Self-managed system and Autonomic Computing, have been proposed. This paper describes an architectural approach for autonomic systems, which is based on a three-layered model. In the uppermost layer, the planning function, which is an important part of this model, has to efficiently make an effective sequence of operational services to satisfy goals. In this paper, we propose an automated planning algorithm using hierarchy planning technique. Our planner composes the sequence of operational services in the most abstract space, and then it successively embodies the detail of them. The critical values, which determine the abstract space, are automatically discovered from the knowledge of operational services. We also present the experimental results to show the effectiveness of our method.
    抄録全体を表示
  • 川村 隆浩, 沈 偉, 中川 博之, 田原 康之, 大須賀 昭彦
    26 巻 (2011) 1 号 p. 116-126
    公開日: 2011/01/06
    ジャーナル フリー
    Ontology-enabled services are rapidly increasing in the Web. However, those are sort of ``lighter'' ontologies compared with ontologies used in design and diagnosis. In this paper, we propose ONTOMO that enables internet users to take part in building those ontologies. ONTOMO is designed not for ontology experts, but for general users of the light-weight ontology. So we focused on easy-use, no installation, and cooperative work environment. Also, it has an agent function which recommends instances and properties belong to ontology classes to boost the users' input. In the recommendation agent, we built our own proper noun extraction mechanism based on bootstrapping. Furthermore, ONTOMO provides a ontology-based blog search as a sample application to motivate the users' ontology building. After the ONTOMO overview, we present the instance and property recommendation agent with experimental evaluation.
    抄録全体を表示
  • 山本 学
    26 巻 (2011) 1 号 p. 127-135
    公開日: 2011/01/06
    ジャーナル フリー
    Recently, highly extensible IT systems are required. It is said that an agent programming model provides high efficiency on application development, however, modification of both data structure and message handler program of agents are needed to add a new service. The issue is the same as the issue that other programming models have. IBM Agent Framework which we developed provides a highly extensible capability. By using the framework, a new service can be added without modifying existing data structure and programming code. In this paper, we introduce the capability and explain how the framework provides the capability. We also discuss why the framework can provide the capability.
    抄録全体を表示
  • 馬場 里美, 岩崎 敦, 横尾 真, Marius C. Silaghi, 平山 勝敏, 松井 俊浩
    26 巻 (2011) 1 号 p. 136-146
    公開日: 2011/01/06
    ジャーナル フリー
    In this paper, we extend the traditional formalization of distributed constraint satisfaction problems (DisCSP) to a quantified DisCSP. A quantified DisCSP includes several universally quantified variables, while all of the variables in a traditional DisCSP are existentially quantified. A universally quantified variable represents a choice of nature or an adversary. A quantified DisCSP formalizes a situation where a team of agents is trying to make a robust plan against nature or an adversary. In this paper, we present the formalization of such a quantified DisCSP and develop an algorithm for solving it by generalizing the asynchronous backtracking algorithm used for solving a DisCSP. In this algorithm, agents communicate a value assignment called a good in addition to the nogood used in asynchronous backtracking. Interestingly, the procedures executed by an adversarial/cooperative agent for good/nogood are completely symmetrical. Furthermore, we develop a method that improves this basic algorithm. Experimental evaluation results illustrate that we observe an easy-hard-easy transition by changing the tightness of the constraints, while very loose problem instances are relatively hard. The modification of the basic algorithm is also effective and reduces the number of cycles by approximately 25% for the hardest problem instances.
    抄録全体を表示
  • 藤田 桂英, 伊藤 孝行, Mark Klein
    26 巻 (2011) 1 号 p. 147-155
    公開日: 2011/01/06
    ジャーナル フリー
    Most real-world negotiation involves multiple interdependent issues, which create agent utility functions that are nonlinear. Our research focuses on developing algorithms that enable this kind of negotiation. We present a novel bidding-based negotiation protocol that addresses the excessively high failure rates that existing approaches face when applied to highly complex nonlinear utility functions. This protocol works by using issue dependency information as follows. First, agents generate an interdependency graph by analyzing the agent's constraints. Second, a mediator identifies issue-groups based on the agents' interdependency graphs. Third, agents generate bids that are divided into these issue-groups. Finally, the mediator identifies the winning contract by finding the best combinations of bids in each issue-group. In this paper, we demonstrate that our proposed protocol is highly scalable when compared to previous efforts in a more realistic experimental setting.
    抄録全体を表示
  • 高田 司郎, 新出 尚之, 藤田 恵
    26 巻 (2011) 1 号 p. 156-165
    公開日: 2011/01/06
    ジャーナル フリー
    TOMATOes is an extension of BDI logic, which introduced probabilistic state transitions and fix-point operators. Using TOMATOes, we can strictly describe and infer various properties of rational agents with those extended notions. In this paper, we give a detailed explanation of modeling of reinforcement learning with the Kripke structure used in TOMATOes, called BDI structure, and the description of transaction graph with policy using TOMATOes. In addition, we give some issues on rational agents for practical reasoning with the description of transaction graph using TOMATOes.
    抄録全体を表示
  • グェン ミンティ, 川村 隆浩, 中川 博之, 田原 康之, 大須賀 昭彦
    26 巻 (2011) 1 号 p. 166-178
    公開日: 2011/01/06
    ジャーナル フリー
    In our definition, human activity can be expressed by five basic attributes: actor, action, object, time and location. The goal of this paper is describe a method to automatically extract all of the basic attributes and the transition between activities derived from sentences in Japanese web pages. However, previous work had some limitations, such as high setup costs, inability to extract all attributes, limitation on the types of sentences that can be handled, and insufficient consideration interdependency among attributes. To resolve these problems, this paper proposes a novel approach that uses conditional random fields and self-supervised learning. Given a small corpus sample as input, it automatically makes its own training data and a feature model. Based on the feature model, it automatically extracts all of the attributes and the transition between the activities in each sentence retrieved from the Web corpus. This approach treats activity extraction as a sequence labeling problem, and has advantages such as domain-independence, scalability, and does not require any human input. Since it is unnecessary to fix the number of elements in a tuple, this approach can extract all of the basic attributes and the transition between activities by making only a single pass. Additionally, by converting to simpler sentences, the approach can deal with complex sentences retrieved from the Web. In an experiment, this approach achieves high precision (activity: 88.9%, attributes: over 90%, transition: 87.5%).
    抄録全体を表示
  • 上田 俊, 岩崎 敦, 横尾 真, Marius C. Silaghi, 平山 勝敏, 松井 俊浩
    26 巻 (2011) 1 号 p. 179-189
    公開日: 2011/01/06
    ジャーナル フリー
    Forming effective coalitions is a major research challenge in AI and multi-agent systems. Coalition Structure Generation (CSG) involves partitioning a set of agents into coalitions so that social surplus (the sum of the rewards of all coalitions) is maximized. A partition is called a coalition structure (CS). In traditional works, the value of a coalition is given by a black box function called a characteristic function. In this paper, we propose a novel formalization of CSG, i.e., we assume that the value of a characteristic function is given by an optimal solution of a distributed constraint optimization problem (DCOP) among the agents of a coalition. A DCOP is a popular approach for modeling cooperative agents, since it is quite general and can formalize various application problems in MAS. At first glance, this approach sounds like a very bad idea considering the computational costs, since we need to solve an NP-hard problem just to obtain the value of a single coalition. To optimally solve a CSG, we might need to solve O(2n) DCOP problem instances, where n is the number of agents. However, quite surprisingly, we show that an approximation algorithm, whose computational cost is about the same as solving just one DCOP, can find a CS whose social surplus is at least max(2/n, 1/(w*+1)) of the optimal CS, where w* is the tree width of a constraint graph. Furthermore, we can generalize this approximation algorithm with a parameter k, i.e., the generalized algorithm can find a CS whose social surplus is at least max(2k/n, k/(w*+1)) of the optimal CS by exploring more search space. These results illustrate that the locality of interactions among agents, which is explicitly modeled in the DCOP formalization, is quite useful in developing efficient CSG algorithms with quality guarantees.
    抄録全体を表示
  • 藤田 宏介, 西野 成昭, 佐藤 勇気, 上田 完次, 淺間 一
    26 巻 (2011) 1 号 p. 190-198
    公開日: 2011/01/06
    ジャーナル フリー
    In service industries, service providers offer various reward programs to consumers with the aim to build customer loyalty and increase sales. In this research, we model membership services as club goods and analyze the reward programs. Club goods are defined as excludable and nonrival public goods. We conduct equilibrium analysis, experiments with human subjects and multi-agent simulation. In theoretical equilibrium, all consumer players become a member. But the results of the experiments indicate that provider players do not properly set the entry fee and the service price for members and consumer players sometimes make irrational decisions of membership entry. Moreover, we elucidate subjects' behavior mechanism of the membership entry by simulations. Results of simulations suggest that the subjects' behavior might be based on the value of how many times they should use services in order to recover payment of the entry fee.
    抄録全体を表示
  • 桂木 敦史, 櫻井 祐子, 岩崎 敦, 横尾 真
    26 巻 (2011) 1 号 p. 199-207
    公開日: 2011/01/06
    ジャーナル フリー
    This paper provides a numerical analysis of Bayesian Nash equilibrium in first-price combinatorial auctions, where participants/agents can use false-name bids. False-name bids is ones submitted by a single agent which uses multiple fictitious names, such as multiple e-mail addresses. It is well-known that even the celebrated Vickrey-Clarke-Groves (VCG) mechanism is influenced by the false-name bids. However, it is not so far investigated how false-name bids affects outcomes of first-price combinatorial auctions, which are widely used in realistic settings. This paper shed a light on the effect of false-name bids in first-price combinatorial auctions, by utilizing Bayesian Nash equilibrium concept via theoretical and numerical analysis.
    抄録全体を表示
  • 八木 勲, 水田 孝信, 和泉 潔
    26 巻 (2011) 1 号 p. 208-216
    公開日: 2011/01/06
    ジャーナル フリー
    Since the subprime mortgage crisis in the United Sates, stock markets around the world have crashed, revealing their instability. To stem the decline in stock prices, short-selling regulations have been implemented in many markets. However, their effectiveness remains unclear. In this paper, we discuss the effectiveness of short-selling regulation using artificial markets. An artificial market that is an agent-based model of financial markets is useful to observe the market mechanism. That is, it is effective for analyzing causal relationship between the behaviors of market participants and the transition of market price. We constructed an artificial market that allows short-selling and an artificial market with short-selling regulation and have observed the stock prices in both of these markets. We have demonstrated that our artificial market had some properties of actual markets. We found that the market in which short-selling was allowed was more stable than the market with short-selling regulation, and a bubble emerged in the regulated market. We evaluated the values of assets of agents who used three trading strategies, specifically, these agents were fundamentalists, chartists, and noise traders. The fundamentalists had the best performance among the three types of agents.
    抄録全体を表示
  • 櫻井 祐子, 横尾 真
    26 巻 (2011) 1 号 p. 217-227
    公開日: 2011/01/06
    ジャーナル フリー
    We develop strategy-proof redistribution mechanisms when bidders have budget limitations. The budget limitation is one of realistic constraints for bidders. There have been several redistribution mechanisms under the assumption of quasi-linear utility functions. As a class of strategy-proof redistribution mechanisms, the partition mechanism has been proposed that divides the bidders into two partitions. Furthermore, in multi-unit auction mechanism for budget constrained bidders, a strategy-proof partition mechanism has been developed that can calculate an appropriate threshold price by using bidding information including gross utility and budget limitation.

    In this paper, by integrating these techniques, we study partition mechanisms that guarantee strategy-proof and budget balance for budget constrained bidders. There exist the flows of information related to the bidders, redistribution payments, and unsold goods among partitions. We provide a condition on the flows so that a partition mechanism can satisfy strategy-proofness as long as the mechanism applied in each partition is strategy-proof. Furthermore, we develop the three mechanisms as examples that satisfy this condition.

    抄録全体を表示
  • 平田 紀史, 白松 俊, 大囿 忠親, 新谷 虎松
    26 巻 (2011) 1 号 p. 228-236
    公開日: 2011/01/06
    ジャーナル フリー
    We propose a system to offer better understanding of news articles on the Web by arranging events. To understand an article, it is necessary to consider background knowledge, details of the article, and meaning of the words. We aim to provide with a better understanding of news articles according to news articles' background by event arrangement. An event arrangement is a graph of related events. We believe that it is difficult to read and understand a topic without knowledge of related events. Arranging events by considering user's interests is necessary to support understanding of the news because each user's interests are different. The system deals with that issue by interaction between user's input and the system output. Processing time and user's interest are important to achieve our goal. The system reduces the processing time by restriction of the processing range using user's input. Event arrangement according to user interest is realized by iterating over states of event presentation and user selection. The experimental results using actual news articles show that the proposed system is effective to detect useful events for understanding news articles.
    抄録全体を表示
一般論文
原著論文
  • 柴田 博仁, 大村 賢悟
    26 巻 (2011) 1 号 p. 237-247
    公開日: 2011/01/06
    ジャーナル フリー
    When we perform tasks in computers, we usually need to refer multiple documents of multiple applications to perform each task. Additionally we usually work in computers with switching multiple tasks consisting of multiple documents in parallel. We propose an extended window system called Docking Window Framework to support such multitasking situations in computers. The proposed system enables to construct a workspace consisting of multiple windows for each task and to easily switch workspaces. The characteristics of our system are (1) to provide docking window interface to easily construct workspaces, (2) to support operating multiple windows simultaneously, (3) to provide a tile layout of windows to reduce the overhead of window operations, and (4) to support saving and reconstructing workspaces. In this paper, we focused on the above first characteristics and carried out a experiment to evaluate the docking window interface. In the task of window arrangement, our system was 20.9 to 23.4% faster than a traditional window system. This indicates that our system provides effective framework to construct workspaces of multiple windows in comparison with previous proposals to support multitasking as well as traditional window systems.
    抄録全体を表示
  • 浅野 優, 田中 譲
    26 巻 (2011) 1 号 p. 248-261
    公開日: 2011/01/06
    ジャーナル フリー
    This paper describes a new framework for querying the Semantic Web using a rich vocabulary. This framework consists of two mechanisms; one for building a rich vocabulary based on lexicographic semantics, and the other for evaluating queries using such a vocabulary. A vocabulary built by the former mechanism has the following two features: (a) its richness because of its expandability and (b) the lexicographic-semantic definition of its words. Query expressions using such a rich vocabulary satisfy the following two properties: (c) no need to use nested query structures, and (d) no need to use variables. In our framework, a new word, i.e., a derived word, can be defined as a character string label given to an expression that combines already defined words with operators. This expression, or phrase, works as a lexicographic definition of this derived word. Each vocabulary consists of basic words and derived words. A lexicon of a vocabulary denotes a set of lexicographic definitions of all of its derived words. Once someone defines a lexicon of a large vocabulary with all of its basic words being mapped to an ontology of the Semantic Web, users can query this Semantic Web using this vocabulary. The same lexicon can be reused for the Semantic Web that has a different ontology if all of its basic words are newly mapped to its ontology. Use of a rich vocabulary in querying a Semantic Web simplifies the query sentence structure and removes the necessity of using variables from each query, which makes it much easier for users to query the Semantic Web. This framework provides query evaluation rules based on the proposed lexicographic semantics, which guarantees that each query using such a rich vocabulary is correctly evaluated over the underlying Semantic Web.
    抄録全体を表示
  • 柴山 直樹, 中川 裕志
    26 巻 (2011) 1 号 p. 262-272
    公開日: 2011/01/06
    ジャーナル フリー
    probabilistic Latent Semantic Indexing (pLSI) is a fundamental method for the analysis of text and related resources which is based on a simple statistical model. This method has high extendibility and scalability due to its simplicity. pLSI is also known as matrix factorization method such as Singular Value Decomposition(SVD) or Non-negative Matrix Factorization. Using pLSI, three matrices which include one diagonal matrix as SVD are achieved. The diagonal elements of this diagonal matrix represent singular values in SVD. However it is not entirely clear what the diagonal matrix of pLSI represents. Then it is also unclear whether the diagonalization constraint is necessary in pLSI.

    This question is the starting point of this paper. To make an answer for this question, we demonstrated that introducing off-diagonal elements to singular value matrix in pLSI is equal to permitting joint probability between different hidden variables. Although permitting joint probability in pLSI does not lose scalability and simplicity, our experiments demonstrated that this extension showed tolerance for over-learning and over-fitting problems.

    抄録全体を表示
  • 金山 博, 那須川 哲哉, 渡辺 日出雄
    26 巻 (2011) 1 号 p. 273-283
    公開日: 2011/01/06
    ジャーナル フリー
    Sentiment analysis is a task to extract and organize authors' evaluations and opinions on focal subjects by analyzing massive amounts of text. This paper proposes a model of tree transfer from a syntactic tree to a set of semantic representations of sentiments. The method is based on deep syntactic and semantic information so that the outputs have suitable features for sentiment analysis applications: (1) to accurately detect the sentiment and its polarity and (2) to aggregate utterances which convey same or similar opinions. The proposed model can be designed analogously to a transfer-based method for machine translation, thus we can reuse several syntactic and semantic operations, such as combination of syntactic subtrees, case analysis of verb phrases and word sense disambiguation, and also several types of syntactic patterns. The experiments on Japanese sentiment extraction show that we acquired the sentiment expression in high-precision, the representation forms were informative than the naive ways of surface extraction and we can develop such a desirable sentiment extraction engine in a systematic way.
    抄録全体を表示
  • 三好 裕樹, 尾崎 知伸, 江口 浩二, 大川 剛直
    26 巻 (2011) 1 号 p. 284-296
    公開日: 2011/01/06
    ジャーナル フリー
    In this paper, we discuss pattern mining problems from a single graph whose vertices or edges contain a set of numerical attributes. Several networks can be naturally represented in this kind of complex graph. A typical example is a social network whose vertex corresponds to a person with some numerical attributes such as age, salary and so on. Another example is a communication network whose edge represents a communication between devices. We can associate the numbers of communications per certain time period to each edge. For these kinds of complex graphs, it is meaningful to consider not only graph strucuture but also the internal informations. Although it can be expected that these kinds of data will increase rapidly, most of current graph mining algorithms do not handle these complex graphs directly. Motivated by the above background, we developed algorithms named FAGV-gSpan and FAGE-gSpan for finding frequent patterns from a single graph with numerical attributes, by effectively combining techniques of graph mining and quantitative itemset mining. The effectiveness of the proposed algorithms was confirmed by experiments using real world datasets.
    抄録全体を表示
  • Takashi Uemura, Daisuke Ikeda, Takuya Kida, Hiroki Arimura
    26 巻 (2011) 1 号 p. 297-306
    公開日: 2011/01/06
    ジャーナル フリー
    In this paper, we study content-based spam detection for spams that are generated by copying a seed document with some random perturbations. We propose an unsupervised detection algorithm based on an entropy-like measure called document complexity, which reflects how many similar documents exist in the input collection of documents. As the document complexity, however, is an ideal measure like Kolmogorov complexity, we substitute an estimated occurrence probability of each document for its complexity. We also present an efficient algorithm that estimates the probabilities of all documents in the collection in linear time to its total length. Experimental results showed that our algorithm especially works well for word salad spams, which are believed to be difficult to detect automatically.
    抄録全体を表示
    • |<
    • <
    • 1
    • >
    • >|
feedback
Top