nach oben

2013 | Buch

Agents and Data Mining Interaction

8th International Workshop, ADMI 2012, Valencia, Spain, June 4-5, 2012, Revised Selected Papers

herausgegeben von: Longbing Cao, Yifeng Zeng, Andreas L. Symeonidis, Vladimir I. Gorodetsky, Philip S. Yu, Munindar P Singh

Verlag: Springer Berlin Heidelberg

Buchreihe : Lecture Notes in Computer Science

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Einloggen, um Zugang zu erhalten

Über dieses Buch

This book constitutes the thoroughly refereed post-workshop proceedings of the 8th International Workshop on Agents and Data Mining Interaction, ADMI 2012, held in Valencia, Spain, in June 2012. The 16 revised full papers were carefully reviewed and selected from numerous submissions. The papers are organized in topical sections on agents for data mining, data mining for agents, and agent mining applications.

Inhaltsverzeichnis

Frontmatter

Invited Talks

Frontmatter

Organizational Control for Data Mining with Large Numbers of Agents

Abstract

Over the last few years, my research group has begun exploring the issues involved in learning when there are hundreds to thousands of agents. We have been using the idea of organization control as a low overhead way of coordinating the learning of such large agent collectives. In this lecture, the results of this research will be discussed and its relationship to issues in distributed data mining.

Victor Lesser

Competitive Benchmarking: Lessons Learned from the Trading Agent Competition

Abstract

Many important developments in artificial intelligence have been stimulated by organized competitions that tackle interesting, difficult challenge problems, such as chess, robot soccer, poker, robot navigation, stock trading, and others. Economics and artificial intelligence share a strong focus on rational behavior. Yet the real-time demands of many domains do not lend hemselves to traditional assumptions of rationality. This is the case in many trading environments, where self-interested entities need to operate subject to limited time and information. With the web mediating an ever broader range of transactions and opening the door for participants to concurrently trade across multiple markets, there is a growing need for technologies that empower participants to rapidly evaluate very large numbers of alternatives in the face of constantly changing market conditions. AI and machine-learning techniques, including neural networks and genetic algorithms, are already routinely used in support of automated trading scenarios. Yet, the deployment of these technologies remains limited, and their proprietary nature precludes the type of open benchmarking that is critical for further scientific progress.

The Trading Agent Competition was conceived to provide a platform for study of agent behavior in competitive economic environments. Research teams from around the world develop agents for these environments. During annual competitions, they are tested against each other in simulated market environments. Results can be mined for information on agent behaviors, and their effects on agent performance, market conditions, and the performance and behavior of competing agents. After each competition, competing agents are made available for offline research. We will discuss results from various competitions (Travel, Supply-Chain Management, Market Design, Sponsored Search, and Power Markets).

Wolfgang Ketter

Agents for Data Mining

Frontmatter

Supporting Agent-Oriented Software Engineering for Data Mining Enhanced Agent Development

Abstract

The emergence of Multi-Agent systems as a software paradigm that most suitably fits all types of problems and architectures is already experiencing significant revisions. A more consistent approach on agent programming, and the adoption of Software Engineering standards has indicated the pros and cons of Agent Technology and has limited the scope of the, once considered, programming ‘panacea’. Nowadays, the most active area of agent development is by far that of intelligent agent systems, where learning, adaptation, and knowledge extraction are at the core of the related research effort. Discussing knowledge extraction, data mining, once infamous for its application on bank processing and intelligence agencies, has become an unmatched enabling technology for intelligent systems. Naturally enough, a fruitful synergy of the aforementioned technologies has already been proposed that would combine the benefits of both worlds and would offer computer scientists with new tools in their effort to build more sophisticated software systems. Current work discusses Agent Academy, an agent toolkit that supports: a) rapid agent application development and, b) dynamic incorporation of knowledge extracted by the use of data mining techniques into agent behaviors in an as much untroubled manner as possible.

Andreas L. Symeonidis, Panagiotis Toulis, Pericles A. Mitkas

Role-Based Management and Matchmaking in Data-Mining Multi-Agent Systems

Abstract

We present an application of concepts of agent, role and group to the hybrid intelligence data-mining tasks. The computational MAS model is formalized in axioms of description logic. Two key functionalities — matchmaking and correctness verification in the MAS — are provided by the role model together with reasoning techniques which are embodied in specific ontology agent. Apart from a simple computational MAS scenario, other configurations such as pre-processing, meta-learning, or ensemble methods are dealt with.

Ondřej Kazík, Roman Neruda

Incentivizing Cooperation in P2P File Sharing

Indirect Interaction as an Incentive to Seed

Abstract

The fundamental problem with P2P networks is that quality of service depends on altruistic resource sharing by participating peers. Many peers freeride on the generosity of others. Current solutions like sharing ratio enforcement and reputation systems are complex, exploitable, inaccurate or unfair at times. The need to design scalable mechanisms that incentivize cooperation is evident. We focus on BitTorrent as the most popular P2P file sharing application and introduce an extension which we refer to as the indirect interaction mechanism (IIM). With IIM BitTorrent peers are able to barter pieces of different files (indirect interaction). We provide novel game theoretical models of BitTorrent and the IIM mechanism and demonstrate through analysis and simulations that IIM improves BitTorrent performance. We conclude that IIM is a practical solution to the fundamental problem of incentivizing cooperation in P2P networks.

Arman Noroozian, Mathijs de Weerdt, Cees Witteveen

An Agent Collaboration-Based Data Hierarchical Caching Approach for HD Video Surveillance

Abstract

In the research of networked HD video surveillance, the agent collaboration has been utilized as an emerging solution to collaborative caching in order to achieve effective adaption among the front-end HD video capture, the network data transmission and the data management for lossless video storage and complete playback. However, the cluster characteristic of various caches embedded in the IP camera, the network proxy server and the data management server, essentially contain important knowledge. How to utilize the cache clustering for collaborative stream controlling is still an open problem. In this paper, we propose an agent collaboration-based 3-level caching (AC3Caching) model, in which a cache storage space-based AP clustering mechanism is developed for fast grouping of “similar” caches on different levels. Furthermore, based on the cache cluster, transmission planning is designed based on the agent collaboration and reasoning. The experimental evaluations demonstrate the capability of the proposed approach.

Wenjia Niu, Xinghua Yang, Gang Li, Endong Tong, Hui Tang, Song Ci

System Modeling of a Smart-Home Healthy Lifestyle Assistant

Abstract

A system modeling is presented for a Smart-home Healthy Lifestyle Assistant System (SHLAS), covering healthy lifestyle promotion by intelligently collecting and analyzing context information, executing control instruction and suggesting health plans for users. SHLAS is Multi-agent based. Each agent has three levels: the Goal Layer has business rules for representing agent goals; the Strategy Layer provides technical rules and processes for guiding how the agent reacts to events; the Component Layer is made up of components, some components are called by technical rules and processes in the Strategy Layer, some others are used for communicating with third party systems. This agent framework enables the customizability of agents in SHLAS. We also introduce an Ontology-based domain knowledge and context model to capture and represent the agents, and agent behavior which provides agents with reasoning ability. SHLAS helps users with healthy lifestyle promotion by tracking and analyzing their behaviors, and recommending health plans. The paper closes with an empirical evaluation of the approach from the point of view of customizability.

Xinhua Zhu, Yaxin Yu, Yuming Ou, Dan Luo, Chengqi Zhang, Jiahang Chen

Data Mining for Agents

Frontmatter

An Optimization Approach to Believable Behavior in Computer Games

Abstract

Many artificial intelligence techniques have been developed to construct intelligent non-player characters (NPCs) in computer games. As games are gradually becoming an integral part of our life, they require human-like NPCs that shall exhibit believable behavior in the game-play. In this paper, we present an optimization approach to designing believable behavior models for NPCs. We quantify the notion of believability using a multi-objective function, and subsequently convert the achieving of believable behavior into one function optimization problem. We compute its analytical solutions and demonstrate the performance in a practical game.

Yifeng Zeng, Hua Mao, Fan Yang, Jian Luo

Discovering Frequent Patterns to Bootstrap Trust

Abstract

When a new agent enters to an open multiagent system, bootstrapping its trust becomes a challenge because of the lack of any direct or reputational evidence. To get around this problem, existing approaches assume the same a priori trust for all newcomers. However, assuming the same a priori trust for all agents may lead to other problems like whitewashing. In this paper, we leverage graph mining and knowledge representation to estimate a priori trust for agents. For this purpose, our approach first discovers significant patterns that may be used to characterise trustworthy and untrustworthy agents. Then, these patterns are used as features to train a regression model to estimate trustworthiness. Lastly, a priori trust for newcomers are estimated using the discovered features based on the trained model. Through extensive simulations, we have showed that the proposed approach significantly outperforms existing approaches.

Murat Sensoy, Burcu Yilmaz, Timothy J. Norman

Subjectivity and Objectivity of Trust

Abstract

Trust plays an important role in the fields of Distributed Artificial Intelligence (DAI) and Multi-agent Systems (MAS), which provides a more effective way to reduce complexity in condition of increasing social complexity. Although a number of computational issues about trust have been studied, there has to date been little attempt to investigate the differences between Objective Trust (OT) and Subjective Trust (ST). In this paper, we will rectify this omission. Particularly, we study the relationship between OT and ST, and propose Transitive Trust (TT) based on ST. We show that, differing with OT, ST is related to preferences of agents. We propose three rules to form trust framework, and give an example to illustrate the process of trust formation. We finally characterize some useful properties of OT and ST.

Xiangrong Tong, Wei Zhang, Yu Long, Houkuan Huang

KNN-Based Clustering for Improving Social Recommender Systems

Abstract

Clustering is useful in tag based recommenders to reduce sparsity of data and by doing so to improve also accuracy of recommendation. Strategy for the selection of tags for clusters has an impact on the accuracy. In this paper we propose a KNN based approach for ranking tag neighbors for tag selection. We study the approach in comparison to several baselines by using two datasets in different domains. We show, that in both cases the approach outperforms the compared approaches.

Rong Pan, Peter Dolog, Guandong Xu

A Probabilistic Model Based on Uncertainty for Data Clustering

Abstract

Recently, all kinds of data in real-life have exploded in an unbelievable way. In order to manage these data, dataspace has been becoming a universal platform, which contains various kinds of data, such as unstructured data, semi-structured data and structured data. But how to cluster these data in dataspace in an efficient and accurate way to help the user manage and explore them is still an intractable problem. In the previous work, the uncertain relationship between term and topic is not considered sufficiently. There are many techniques to handle this problem and probability theory provides an effective way to deal with the uncertainty of clustering. As a result, we proposed a novel probability model based on topic terms, i.e., Probabilistic Term Similarity Model (PTSM) to tackle the uncertainty between term and topic. In this model, not only terms from various data but also structure information of semi-structured and structured data are considered. Each term is assigned a probability indicating how relevant it is to the topic. Then, according to the probability for each term, a probabilistic matrix is established for clustering various data. At last, extensive experiment results show that the clustering method based on this probabilistic model has excellent performance and outperforms some other classical algorithms.

Yaxin Yu, Xinhua Zhu, Miao Li, Guoren Wang, Dan Luo

Following Human Mobility Using Tweets

Abstract

The availability of location-based agent data is growing rapidly, enabling new research into the behavior patterns of such agents in space and time. Previously, such analysis was limited to either small experiments with GPS-equipped agents, or proprietary datasets of human cell phone users that cannot be disseminated across the academic community for followup studies. In this paper, we study the movement patterns of Twitter users in London, Los Angeles, and Tokyo. We cluster these agents by their movement patterns across space and time. We also show that it is possible to infer part of the underlying transportation net- work from Tweets alone, and uncover interesting differences between the behaviors exhibited by users across these three cities.

Mahdi Azmandian, Karan Singh, Ben Gelsey, Yu-Han Chang, Rajiv Maheswaran

Agent Mining Applications

Frontmatter

Agents and Distributed Data Mining in Smart Space: Challenges and Perspectives

Abstract

Smart space is a distributed ambient environment with existing, inside it, dynamic set of inhabitants (living and nonliving) solving various own and common tasks. The mission of smart space is to provide, for its inhabitants, with context–dependent information, communication, services, reminders and personalized recommendations in a user–friendly mode where and when needed in ubiquitous and unobtrusive style. The smart space R&D uses large diversity of models, frameworks, and technologies and their integration is the first challenging smart space problem. Another challenge is caused by the necessity to process huge volumes of heterogeneous information perceived by distributed sensors in adaptive, self–organizing, learnable, and efficient style. The paper analyses these challenges and emphasizes an important role of the technology integrating agent and data mining to overcome both these challenges.

Vladimir Gorodetsky

Agent-Mining of Grid Log-Files: A Case Study

Abstract

Grid monitoring requires analysis of large amounts of log files across multiple domains. An approach is described for automated extraction of job-flow information from large computer grids, using software agents and genetic computation. A prototype was created as a first step towards communities of agents that will collaborate to learn log-file structures and exchange knowledge across organizational domains.

Arjan J. R. Stoter, Simon Dalmolen, Wico Mulder

A Proficient and Dynamic Bidding Agent for Online Auctions

Abstract

E-consumers face biggest challenge of opting for the best bidding strategies for competing in an environment of multiple and simultaneous online auctions for same or similar items. It becomes very complicated for the bidders to make decisions of selecting which auction to participate in, place single or multiple bids, early or late bidding and how much to bid. In this paper, we present the design of an autonomous dynamic bidding agent (ADBA) that makes these decisions on behalf of the buyers according to their bidding behaviors. The agent develops a comprehensive method for initial price prediction and an integrated model for bid forecasting. The initial price prediction method selects an auction to participate in and then predicts its closing price (initial price). Then the bid forecasting model forecasts the bid amount by designing different bidding strategies followed by the late bidders. The experimental results demonstrated improved initial price prediction outcomes by proposing a clustering based approach. Also, the results show the proficiency of the bidding strategies amongst the late bidders with desire for bargain.

Preetinder Kaur, Madhu Goyal, Jie Lu

Trading Strategy Based Portfolio Selection for Actionable Trading Agents

Abstract

Trading agents are very useful for supporting investors in making decisions in financial markets, but the existing trading agent research focuses on simulation on artificial data. This leads to limitations in its usefulness. As for investors, how trading agents help them manipulate their assets according to their risk appetite and thus obtain a higher return is a big issue. Portfolio optimization is an approach used by many researchers to resolve this issue, but the focus is mainly on developing more accurate mathematical estimation methods, and overlooks an important factor: trading strategy. Since the global financial crisis added uncertainty to financial markets, there is an increasing demand for trading agents to be more active in providing trading strategies that will better capture trading opportunities. In this paper, we propose a new approach, namely trading strategy based portfolio selection, by which trading agents combine assets and their corresponding trading strategies to construct new portfolios, following which, trading agents can help investors to obtain the optimal weights for their portfolios according to their risk appetite. We use historical data to test our approach, the results show that it can help investors make more profit according to their risk tolerance by selecting the best portfolio in real financial markets.

Wei Cao, Cheng Wang, Longbing Cao

On the Need of New Methods to Mine Electrodermal Activity in Emotion-Centered Studies

Abstract

Monitoring the electrodermal activity is increasingly accomplished in agent-based experimental settings as the skin is believed to be the only organ to react only to the sympathetic nervous system. This physiological signal has the potential to reveal paths that lead to excitement, attention, arousal and anxiety. However, electrodermal analysis has been driven by simple feature-extraction, instead of using expressive models that consider a more flexible behavior of the signal for improved emotion recognition. This paper proposes a novel approach centered on sequential patterns to classify the signal into a set of key emotional states. The approach combines SAX for pre-processing the signal and hidden Markov models. This approach was tested over a collected sample of signals using Affectiva-QSensor. An extensive human-to-human and human-to-robot experimental setting is under development for further validation and characterization of emotion-centered patterns.

Rui Henriques, Ana Paiva, Cláudia Antunes

Backmatter

Titel: Agents and Data Mining Interaction
herausgegeben von: Longbing Cao
Yifeng Zeng
Andreas L. Symeonidis
Vladimir I. Gorodetsky
Philip S. Yu
Munindar P Singh
Verlag: Springer Berlin Heidelberg
Electronic ISBN: 978-3-642-36288-0
Print ISBN: 978-3-642-36287-3
DOI: https://doi.org/10.1007/978-3-642-36288-0