Catálogo de publicaciones - libros

Peer-to-Peer (P2P) networks are gaining increasing popularity in many distributed applications such as file-sharing, network storage, web caching, sear- ching and indexing of relevant documents and P2P network-threat analysis. Many of these applications require scalable analysis of data over a P2P network. This paper starts by offering a brief overview of distributed data mining applications and algorithms for P2P environments. Next it discusses some of the privacy concerns with P2P data mining and points out the problems of existing privacy-preserving multi-party data mining techniques. It further points out that most of the nice assumptions of these existing privacy preserving techniques fall apart in real-life applications of privacy-preserving distributed data mining (PPDM). The paper offers a more realistic formulation of the PPDM problem as a multi-party game and points out some recent results.

This paper deals with the problem of development and implementation of semantic navigation through Web-content. Multi-agent architecture of a solution for Semantic Web and innovative services are presented. In the context of the proposed solution Web mining is carried out by special OntosMiner agents, which provide the ontology-driven processing of multilingual text collections on the basis of the special kind of content extraction technologies. First evaluation results of the presented solution are discussed as well.

We believe that intelligent information agents will represent their users interest in electronic marketplaces and other forums to trade, exchange, share, identify, and locate goods and services. Such information worlds will present unforeseen opportunities as well as challenges that can be best addressed by robust, self-sustaining agent communities. An agent community is a stable, adaptive group of self-interested agents that share common resources and must coordinate their efforts to effectively develop, utilize and nurture group resources and organization. More specifically, agents will need mechanisms to benefit from complementary expertise in the group, pool together resources to meet new demands and exploit transient opportunities, negotiate fair settlements, develop norms to facilitate coordination, exchange help and transfer knowledge between peers, secure the community against intruders, and learn to collaborate effectively. In this talk, I will summarize some of our research results on trust-based computing, negotiation, and learning that will enable intelligent agents to develop and sustain robust, adaptive, and successful agent communities.

In order to investigate human information processing mechanism systematically, Web intelligence (WI) based portal techniques are required for brain data measurement, management and analysis. Building a brain informatics portal is, in fact, to develop a data mining grid centric multi-layer grid system on the Wisdom Web, on which various data mining agents are deployed, for multi-aspect data analysis. We propose an approach for collecting, modeling, transforming, managing, and mining multiple human brain data obtained from systematic fMRI/EEG experiments. The proposed approach provides a new way in Brain Informatics (BI) for automatic analysis and understanding of human brain data to replace human-expert centric visualization. We attempt to change the perspective of cognitive scientists from a single type of experimental data analysis towards a holistic view at a long-term, global field of vision to understand the principle, models and mechanisms of human information processing system.

In the past twenty years, (we mean autonomous agent and multi-agent systems) and (also knowledge discovery) have emerged separately as two of most prominent, dynamic and exciting research areas. In recent years, an increasingly remarkable trend in both areas is . This is driven by not only researcher’s interests, but intrinsic challenges and requirements from both sides, as well as benefits and complementarity to both communities through agent-mining interaction. In this paper, we draw a high-level overview of the agent-mining interaction from the perspective of an emerging area in the scientific family. To promote it as a newly emergent scientific field, we summarize key driving forces, originality, major research directions and respective topics, and the progression of research groups, publications and activities of agent-mining interaction. Both theoretical and application-oriented aspects are addressed. The above investigation shows that the agent-mining interaction is attracting everincreasing attention from both agent and data mining communities. Some complicated challenges in either community may be effectively and efficiently tackled through agent-mining interaction. However, as a new open area, there are many issues waiting for research and development from theoretical, technological and practical perspectives.

As modern applications tend to stretch between large, ever-growing datasets and increasing demand for meaningful content at the user end, more elaborate and sophisticated knowledge extraction technologies are needed. Towards this direction, the inherently contradicting technologies of deductive software agents and inductive data mining have been integrated, in order to address knowledge intensive problems. However, there exists no generalized evaluation methodology for assessing the efficiency of such applications. On the one hand, existing data mining evaluation methods focus only on algorithmic precision, ignoring overall system performance issues. On the other hand, existing systems evaluation techniques are insufficient, as the emergent intelligent behavior of agents introduce unpredictable factors of performance. In this paper, we present a generalized methodology for performance evaluation of intelligent agents that employ knowledge models produced through data mining. The proposed methodology consists of concise steps for selecting appropriate metrics, defining measurement methodologies and aggregating the measured performance indicators into thorough system characterizations. The paper concludes with a demonstration of the proposed methodology to a real world application, in the Supply Chain Management domain.

Over the past few years, Electronic Commerce has become an increasingly central part of the economy. More and more transactions, both from business to consumer and between businesses, are taking place online. Simple fixed cost business transactions are often automated at one or both ends and auctions are mainly conducted by automated auctioneer software. Bid evaluation is a costly optimization problem. It involves the examination of all possible combinations of bids and subsequently the selection of the best combination; based on factors such as price, supplier reliability, schedule risk, etc. If an exhaustive search approach is used to derive the optimum solution, the generation of possible bid combinations will explode combinatorial as the number of bids grow. This paper shows how to build an ant system for autonomous agents. In addition, it assigns customer agents’ services to supplier agents’ bids in the best possible way while considering several factors.

To facilitate automated agent-based software development and to support high effective and low cost selection, adaptation and integration of required functionality extensions into existing multi-agent software systems, we present our semantic modelling approach. It defines a structured annotation process, proposing concepts, techniques and a methodical support for the formal description of static and dynamic semantic information of software agents and their services emerging from heterogeneous environments. This information is organized by description patterns which are – according to our Logic-on-Demand concept – based on a variety of inference mechanisms which offer variability in expressiveness, reasoning power and the required analysis depth for the identification of agent properties and qualities.

This paper describes a new framework using intelligent agents for pattern recognition. Based on Jordan Curve Theorem, a universal classification method called Hyper Surface Classifier (HSC) has been studied since 2002. We propose multi-agents based technology to realize the combination of Hyper Surface Classifiers. Agents can imitate human beings’ group decision to solve problems. We use two types of agents: the classifier training agent and the classifier combining agent. Each classifier training agent is responsible to read a vertical slice of the samples and train the local classifier, while the classifier combining agent is designed to combine the classification results of all the classifier training agents. The key of our method is that the sub-datasets for the classifier training agents are obtained by dividing the features rather than by dividing the sample set in distribution environment. Experimental results show that this method has a preferable performance on high dimensional datasets.

Mobile agent technology offers a new approach to mining information from data sources distributed over the Internet. However, the potential benefits of mobile agents must be weighted against the real security threats. An agent based Distributed Data Mining (DDM) system must cope with exposed data services and insecure communication channels in the Internet to protect the privacy, integrity and availability of agents and distributed resources. In this paper, we present the design of a mobile agent infrastructure for DDM applications and discuss the implementation of security mechanism that has been effectively integrated with the mobile agent virtual machine.