Multi-Agent–Based Simulation of University Email Communities

In this chapter, a study on informal communication network formation in a university environment is presented. The teacher communication network is analyzed through community detection techniques. It is evident that informal communication is an important process that traverses the vertical hierarchical structure of departments and courses in a university environment. A multi-agent model of the case study is presented here, showing the implications of using real data as training sets for multi-agent-based simulations. The influence of the “social neighborhood,” as a mechanism to create assortative networks of contacts without full knowledge of the network, is discussed. It is shown that the radius of this social neighborhood has an effect on the outcome of the network structure and that in a university’s case this distance is relatively small.

Governance Mechanisms in Web 2.0

This chapter documents the findings of research into the governance mechanisms within the distributed on-line community known as Wikipedia. It focuses in particular on the role of normative mechanisms in achieving social self-regulation. A brief history of the Wikipedia is provided. This concentrates on the debate about governance and also considers characteristics of the wiki technology which can be expected to influence governance processes. The empirical findings are then presented. These focus on how Wikipedians use linguistic cues to influence one another on a sample of discussion pages drawn from both controversial and featured articles. Through this analysis a tentative account is provided of the agent-level cognitive mechanisms which appear necessary to explain the apparent behavioural coordination. The findings were to be used as a foundation for the simulation of ‘normative’ behaviour. The account identifies some of the challenges that need to be addressed in such an attempt including a mismatch between the case findings and assumptions used in past attempts to simulate normative behaviour.

Two Informational Complexity Measures in Social Networks and Agent Communities

Several informational complexity measures rely on the notion of stochastic process in order to extract hidden structural properties behind the apparent randomness of information sources. Following an equivalence approach between dynamic relation evolution within a social network and a generic stochastic process two dynamic measures of network complexity are proposed.

Statistical Properties of Social Group Evolution

The authors’ focus is on the general statistical features of the time evolution of communities (also called as modules, clusters or cohesive groups) in large social networks. These structural sub-units can correspond to highly connected circles of friends, families, or professional cliques, which are subject to constant change due to the intense fluctuations in the activity and communication patterns of people. The communities can grow by recruiting new members, or contract by loosing members; two (or more) groups may merge into a single community, while a large enough social group can split into several smaller ones; new communities are born and old ones may disappear. According to our results, the time evolution of social groups containing only a few members and larger communities, e.g., institutions show significant differences.

Towards Learning ‘Self’ and Emotional Knowledge in Social and Cultural Human-Agent Interactions

This article presents research towards the development of a virtual learning environment (VLE) inhabited by intelligent virtual agents (IVAs) and modelling a scenario of inter-cultural interactions. The ultimate aim of this VLE is to allow users to reflect upon and learn about intercultural communication and collaboration. Rather than predefining the interactions among the virtual agents and scripting the possible interactions afforded by this environment, we pursue a bottom-up approach whereby inter-cultural communication emerges from interactions with and among autonomous agents and the user(s). The intelligent virtual agents that are inhabiting this environment are expected to be able to broaden their knowledge about the world and other agents, which may be of different cultural backgrounds, through interactions. This work is part of a collaborative effort within a European research project called eCIRCUS. Specifically, this article focuses on our continuing research concerned with emotional knowledge learning in autobiographic social agents.

A Step-By-Step Implementation of a Multi-Agent Currency Trading System

With this chapter the authors intend to demonstrate the potential practical use of intelligent agents as autonomous financial traders. The authors propose an architecture to be utilized in the creation of this type of agents, consisting of an ensemble of classification and regression models, a case-based reasoning system and an expert system. This architecture was used to implement six intelligent agents, each being responsible for trading one of the following currency pairs with a 6-hour timeframe: CHF/JPY, EUR/CHF, EUR/JPY, EUR/USD, USD/CHF and USD/JPY. These agents simulated trades during an out-of-sample period going from February of 2007 till July of 2010, having all achieved an acceptable performance. However, their strategies resulted in relatively high drawdowns, and much of their profit disappeared once the trading costs were factored into the trading simulation. In order to overcome these problems, they integrated the agents in a multi-agent system, in which agents communicate their decisions to each other before sending the market orders, and work together to eliminate redundant trades. This system averaged out the returns of the agents, thus eliminating much of the risk associated with their individual trading strategies, and also originated considerable savings in trading expenses. Their results seem to vindicate the usefulness of the proposed trading agent architecture, and also demonstrate that there is indeed a place for intelligent agents in financial markets.

Temporally Autonomous Agent Interaction

This work reflects the results of continuing research into “temporally autonomous” multi-agent interaction. Many traditional approaches to modeling multi-agent systems involve synchronizing all agent activity in simulated environments to a single “universal” clock. In other words, agent behavior is regulated by a global timer where all agents act and interact deterministically in time. However, if the objective of any such simulation is to model the behavior of real-world entities, this discrete timing mechanism yields an artificially constrained representation of actual physical agent interaction. In addition to the behavioral autonomy normally associated with agents, simulated agents must also have temporal autonomy in order to interact realistically. Intercommunication should occur without global coordination or synchronization. To this end, a specialized simulation framework is developed. Several simulations are conducted from which data are gathered and we subsequently demonstrate that manipulation of the timing variable amongst interacting agents affects the emergent behaviors of agent populations.

Modeling a Multi-Agents System as a Network

The article presents an empirically oriented investigation on the dynamics of a specific case of a multi-agents system, the stock market. It demonstrates that S&P500 market space can be described using the geometrical and topological characteristics of its dynamics. The authors proposed to measure the coefficient R, an index providing information on the evolution of a manifold describing the dynamics of the market. It indicates the moments of perturbations, proving that the dynamics is driven by shocks and by a structural change. This dynamics has a characteristic dimension, which also allows for a description of its evolution. The consequent description of the market as a network of stocks is useful for the identification of patterns that emerge from multi-agent interaction, and defines our research, as it is derived from a system of measure and it is part of the logic of a defined mathematics.

A Multi-Agent Machine Learning Framework for Intelligent Energy Demand Management

In order to cope with the unpredictability of the energy market and provide rapid response when supply is strained by demand, an emerging technology, called energy demand management, enables appliances to manage and defer their electricity consumption when price soars. Initial experiments with our multi-agent, power load management simulator, showed a marked reduction in energy consumption when price-based constraints were imposed on the system. However, these results also revealed an unforeseen, negative effect: that reducing consumption for a bounded time interval decreases system stability. The reason is that price-driven control synchronizes the energy consumption of individual agents. Hence price, alone, is an insufficient measure to define global goals in a power load management system. In this chapter the authors explore the effectiveness of a multi-objective, system-level goal which combines both price and system stability. The authors apply the commonly known reinforcement learning framework, enabling the energy distribution system to be both cost saving and stable. They test the robustness of their algorithm by applying it to two separate systems, one with indirect feedback and one with direct feedback from local load agents. Results show that their method is not only adaptive to multiple systems, but is also able to find the optimal balance between both system stability and energy cost.

Interactions in Context-Zero

Designing computer interfaces and other technologies that interact with users in adaptive ways that attempt to emulate natural styles of learning is generally difficult. As technology has become common in our daily interactions, adaptive interfaces are key in helping users in many situations. In this chapter the preliminary investigation with the intelligent agent Izbuhska is discussed, along with how it can be used to collect various data from users in an attempt to understand how they perceive the program and “learn” while interacting. Izbushka as a tool will help to generate new ways of understanding and conceptualizing interaction by presenting users with a “zero-context” environment. Izbushka presents users with a unique interface in an attempt to study user interactions that lack traditional metaphors or ontological grounding typical in many computer interfaces. The Izbushka agent is our first step towards filtering our preconceived metaphorical ideas in order to generate new understanding of human-computer interaction.