Modelling the Provenance of Data in Autonomous Systems

Determining the provenance of data, i.e. the process that led to that data, is vital in many disciplines. For example, in science, the process that produced a given result must be demonstrably rigorous for the result to be deemed reliable. A provenance system supports applications in recording adequate documentation about process executions to answer queries regarding provenance, and provides functionality to perform those queries. Several provenance systems are being developed, but all focus on systems in which the components are textitreactive, for example Web Services that act on the basis of a request, job submission system, etc. This limitation means that questions regarding the motives of autonomous actors, or textitagents, in such systems remain unanswerable in the general case. Such questions include: who was ultimately responsible for a given effect, what was their reason for initiating the process and does the effect of a process match what was intended to occur by those initiating the process? In this paper, we address this limitation by integrating two solutions: a generic, re-usable framework for representing the provenance of data in service-oriented architectures and a model for describing the goal-oriented delegation and engagement of agents in multi-agent systems. Using these solutions, we present algorithms to answer common questions regarding responsibility and success of a process and evaluate the approach with a simulated healthcare example.

Norm-based behaviour modification in BDI agents

While there has been much work on developing frameworks and models of norms and normative systems, consideration of the impact of norms on the practical reasoning of agents has attracted less attention. The problem is that traditional agent architectures and their associated languages provide no mechanism to adapt an agent at runtime to norms constraining their behaviour. This is important because if BDI-type agents are to operate in open environments, they need to adapt to changes in the norms that regulate such environments. In response, in this paper we provide a technique to extend BDI agent languages, by enabling them to enact behaviour modification at runtime in response to newly accepted norms. Our solution consists of creating new plans to comply with obligations and suppressing the execution of existing plans that violate prohibitions. We demonstrate the viability of our approach through an implementation of our solution in the AgentSpeak(L) language.

Interactions between normative systems and software cognitive agents. A formalization in temporal modal defeasible logic and its implementation

Sustainable computer systems require some flexibility to adapt to environmental unpredictable changes. A solution lies in autonomous software agents which can adapt autonomously to their environments. Though autonomy allows agents to decide which behavior to adopt, a disadvantage is a lack of control, and as a side effect even untrustworthiness: we want to keep some control over such autonomous agents. How to control autonomous agents while respecting their autonomy? A solution is to regulate agents’ behavior by norms. The normative paradigm makes it possible to control autonomous agents while respecting their autonomy, limiting untrustworthiness and augmenting system compliance. It can also facilitate the design of the system, for example, by regulating the coordination among agents. However, an autonomous agent will follow norms or violate them in some conditions. What are the conditions in which a norm is binding upon an agent? While autonomy is regarded as the driving force behind the normative paradigm, cognitive agents provide a basis for modeling the bindingness of norms. In order to cope with the complexity of the modeling of cognitive agents and normative bindingness, we adopt an intentional stance. Since agents are embedded into a dynamic environment, things may not pass at the same instant. Accordingly, our cognitive model is extended to account for some temporal aspects. Special attention is given to the temporal peculiarities of the legal domain such as, among others, the time in force and the time in efficacy of provisions. Some types of normative modifications are also discussed in the framework. It is noteworthy that our temporal account of legal reasoning is integrated to our commonsense temporal account of cognition. As our intention is to build sustainable reasoning systems running unpredictable environment, we adopt a declarative representation of knowledge. A declarative representation of norms will make it easier to update their system representation, thus facilitating system maintenance; and to improve system transparency, thus easing system governance. Since agents are bounded and are embedded into unpredictable environments, and since conflicts may appear amongst mental states and norms, agent reasoning has to be defeasible, i.e. new pieces of information can invalidate formerly derivable conclusions. In this dissertation, our model is formalized into a non-monotonic logic, namely into a temporal modal defeasible logic, in order to account for the interactions between normative systems and software cognitive agents.

Designing and Programming Organizational Infrastructures for Agents situated in Artifact-based Environments

Actual trends in software development are pushing the need to face a multiplicity of diverse activities and interaction styles characterizing complex and distributed application domains, in such a way that the resulting dynamics exhibits some grade of order, i.e. in terms of evolution of the system and desired equilibrium. Autonomous agents and Multiagent Systems are argued in literature as one of the most immediate approaches for describing such a kind of challenges. Actually, agent research seems to converge towards the definition of renewed abstraction tools aimed at better capturing the new demands of open systems. Besides agents, which are assumed as autonomous entities purposing a series of design objectives, Multiagent Systems account new notions as first-class entities, aimed, above all, at modeling institutional/organizational entities, placed for normative regulation, interaction and teamwork management, as well as environmental entities, placed as resources to further support and regulate agent work. The starting point of this thesis is recognizing that both organizations and environments can be rooted in a unifying perspective. Whereas recent research in agent systems seems to account a set of diverse approaches to specifically face with at least one aspect within the above mentioned, this work aims at proposing a unifying approach where both agents and their organizations can be straightforwardly situated in properly designed working environments. In this line, this work pursues reconciliation of environments with sociality, social interaction with environment based interaction, environmental resources with organizational functionalities with the aim to smoothly integrate the various aspects of complex and situated organizations in a coherent programming approach. Rooted in Agents and Artifacts (A&A) meta-model, which has been recently introduced both in the context of agent oriented software engineering and programming, the thesis promotes the notion of Embodied Organizations, characterized by computational infrastructures attaining a seamless integration between agents, organizations and environmental entities.

Cyber Enigmas? Passive Detection and Pedagogical Agents: Can Students Spot the fake?

This paper presents a study that was undertaken to examine human interaction with a pedagogical agent and the passive and active detection of such agents within a synchronous, online environment. A pedagogical agent is a software application which can provide a human like interaction using a natural language interface. These may be familiar from the smartphone interfaces such as ‘Siri’ or ‘Cortana’, or the virtual online assistants found on some websites, such as ‘Anna’ on the Ikea website. Pedagogical agents are characters on the computer screen with embodied life-like behaviours such as speech, emotions, locomotion, gestures, and movements of the head, the eye, or other parts of the body. The passive detection test is where participants are not primed to the potential presence of a pedagogical agent within the online environment. The active detection test is where participants are primed to the potential presence of a pedagogical agent. The purpose of the study was to examine how people passively detected pedagogical agents that were presenting themselves as humans in an online environment. In order to locate the pedagogical agent in a realistic higher education online environment, problem-based learning online was used. Problem-based learning online provides a focus for discussions and participation, without creating too much artificiality. The findings indicated that the ways in which students positioned the agent tended to influence the interaction between them. One of the key findings was that since the agent was focussed mainly on the pedagogical task this may have hampered interaction with the students, however some of its non-task dialogue did improve students' perceptions of the autonomous agents’ ability to interact with them. It is suggested that future studies explore the differences between the relationships and interactions of learner and pedagogical agent within authentic situations, in order to understand if students' interactions are different between real and virtual mentors in an online setting.

An agent-based approach to modelling driver route choice behaviour under the influence of real-time information

This paper presents an agent-based approach to modelling individual driver behaviour under the influence of real-time traffic information. The driver behaviour models developed in this study are based on a behavioural survey of drivers which was conducted on a congested commuting corridor in Brisbane, Australia. Commuters' responses to travel information were analysed and a number of discrete choice models were developed to determine the factors influencing drivers' behaviour and their propensity to change route and adjust travel patterns. Based on the results obtained from the behavioural survey, the agent behaviour parameters which define driver characteristics, knowledge and preferences were identified and their values determined. A case study implementing a simple agent-based route choice decision model within a microscopic traffic simulation tool is also presented. Driver-vehicle units (DVUs) were modelled as autonomous software components that can each be assigned a set of goals to achieve and a database of knowledge comprising certain beliefs, intentions and preferences concerning the driving task. Each DVU provided route choice decision-making capabilities, based on perception of its environment, that were similar to the described intentions of the driver it represented. The case study clearly demonstrated the feasibility of the approach and the potential to develop more complex driver behavioural dynamics based on the belief-desire-intention agent architecture. (C) 2002 Elsevier Science Ltd. All rights reserved.

Dynamic equilibrium through reinforcement learning

Reinforcement Learning is an area of Machine Learning that deals with how an agent should take actions in an environment such as to maximize the notion of accumulated reward. This type of learning is inspired by the way humans learn and has led to the creation of various algorithms for reinforcement learning. These algorithms focus on the way in which an agent’s behaviour can be improved, assuming independence as to their surroundings. The current work studies the application of reinforcement learning methods to solve the inverted pendulum problem. The importance of the variability of the environment (factors that are external to the agent) on the execution of reinforcement learning agents is studied by using a model that seeks to obtain equilibrium (stability) through dynamism – a Cart-Pole system or inverted pendulum. We sought to improve the behaviour of the autonomous agents by changing the information passed to them, while maintaining the agent’s internal parameters constant (learning rate, discount factors, decay rate, etc.), instead of the classical approach of tuning the agent’s internal parameters. The influence of changes on the state set and the action set on an agent’s capability to solve the Cart-pole problem was studied. We have studied typical behaviour of reinforcement learning agents applied to the classic BOXES model and a new form of characterizing the environment was proposed using the notion of convergence towards a reference value. We demonstrate the gain in performance of this new method applied to a Q-Learning agent.

Sobrevivência: jogo interactivo para estudo de arquitecturas de agentes autónomos

Trabalho de projeto para obtenção do grau de Mestre em Engenharia Informática e de Computadores

A proposal for media component brokerage

This paper describes how MPEG-4 object based video (obv) can be used to allow selected objects to be inserted into the play-out stream to a specific user based on a profile derived for that user. The application scenario described here is for personalized product placement, and considers the value of this application in the current and evolving commercial media distribution market given the huge emphasis media distributors are currently placing on targeted advertising. This level of application of video content requires a sophisticated content description and metadata system (e.g., MPEG-7). The scenario considers the requirement for global libraries to provide the objects to be inserted into the streams. The paper then considers the commercial trading of objects between the libraries, video service providers, advertising agencies and other parties involved in the service. Consequently a brokerage of video objects is proposed based on negotiation and trading using intelligent agents representing the various parties. The proposed Media Brokerage Platform is a multi-agent system structured in two layers. In the top layer, there is a collection of coarse grain agents representing the real world players – the providers and deliverers of media contents and the market regulator profiler – and, in the bottom layer, there is a set of finer grain agents constituting the marketplace – the delegate agents and the market agent. For knowledge representation (domain, strategic and negotiation protocols) we propose a Semantic Web approach based on ontologies. The media components contents should be represented in MPEG-7 and the metadata describing the objects to be traded should follow a specific ontology. The top layer content providers and deliverers are modelled by intelligent autonomous agents that express their will to transact – buy or sell – media components by registering at a service registry. The market regulator profiler creates, according to the selected profile, a market agent, which, in turn, checks the service registry for potential trading partners for a given component and invites them for the marketplace. The subsequent negotiation and actual transaction is performed by delegate agents in accordance with their profiles and the predefined rules of the market.

Plataforma de negociação de componentes multimédia

O objectivo da tese é demonstrar a adequação do paradigma dos mercados electrónicos baseados em agentes para transaccionar objectos multimédia em função do perfil dos espectadores. Esta dissertação descreve o projecto realizado no âmbito da plataforma de personalização de conteúdos em construção. O domínio de aplicação adoptado foi a personalização dos intervalos publicitários difundidos pelos distribuidores de conteúdos multimédia, i.e., pretende-se gerar em tempo útil o alinhamento de anúncios publicitários que melhor se adeqúe ao perfil de um espectador ou de um grupo de espectadores. O projecto focou-se no estudo e selecção das tecnologias de suporte, na concepção da arquitectura e no desenvolvimento de um protótipo que permitisse realizar diversas experiências nomeadamente com diferentes estratégias e tipos de mercado. A arquitectura proposta para a plataforma consiste num sistema multiagente organizado em três camadas que disponibiliza interfaces do tipo serviço Web com o exterior. A camada de topo é constituída por agentes de interface com o exterior. Na camada intermédia encontram-se os agentes autónomos que modelam as entidades produtoras e consumidoras de componentes multimédia assim como a entidade reguladora do mercado. Estes agentes registam-se num serviço de registo próprio onde especificam os componentes multimédia que pretendem negociar. Na camada inferior realiza-se o mercado que é constituído por agentes delegados dos agentes da camada superior. O lançamento do mercado é efectuado através de uma interface e consiste na escolha do tipo de mercado e no tipo de itens a negociar. Este projecto centrou-se na realização da camada do mercado e da parte da camada intermédia de apoio às actividades de negociação no mercado. A negociação é efectuada em relação ao preço da transmissão do anúncio no intervalo em preenchimento. Foram implementados diferentes perfis de negociação com tácticas, incrementos e limites de variação de preço distintos. Em termos de protocolos de negociação, adoptou-se uma variante do Iterated Contract Net – o Fixed Iterated Contract Net. O protótipo resultante foi testado e depurado com sucesso.

Solving Conflicting Beliefs with a Distributed Belief Revision Approach

The ability to solve conflicting beliefs is crucial for multi- agent systems where the information is dynamic, incomplete and dis- tributed over a group of autonomous agents. The proposed distributed belief revision approach consists of a distributed truth maintenance sy- stem and a set of autonomous belief revision methodologies. The agents have partial views and, frequently, hold disparate beliefs which are au- tomatically detected by system’s reason maintenance mechanism. The nature of these conflicts is dynamic and requires adequate methodolo- gies for conflict resolution. The two types of conflicting beliefs addressed in this paper are Context Dependent and Context Independent Conflicts which result, in the first case, from the assignment, by different agents, of opposite belief statuses to the same belief, and, in the latter case, from holding contradictory distinct beliefs. The belief revision methodology for solving Context Independent Con- flicts is, basically, a selection process based on the assessment of the cre- dibility of the opposing belief statuses. The belief revision methodology for solving Context Dependent Conflicts is, essentially, a search process for a consensual alternative based on a “next best” relaxation strategy.

Environmental Decision Support: a Multi-Agent Approach

The environmental management domain is vast and encompasses many identifiable activities: impact assessment, planning, project evaluation, etc. In particular, this paper focusses on the modelling of the project evaluation activity. The environmental decision support system under development aims to provide assistance to project developers in the selection of adequate locations, guaranteeing the compliance with the applicable regulations and the existing development plans as well as satisfying the specified project requirements. The inherent multidisciplinarity features of this activity lead to the adoption of the Multi-Agent paradigm, and, in particular, to the modelling of the involved agencies as a community of cooperative autonomous agents, where each agency contributes with its share of problem solving to the final system’s recommendation. To achieve this behaviour the many conclusions of the individual agencies have to be justifiably accommodated: not only they may differ, but can be interdependent, complementary, irreconcilable, or simply, independent. We propose different solutions (involving both local and global consistency) to support the adequate merge of the distinct perspectives that inevitably arise during this type of decision making.

Machine ethics via logic programming

Machine ethics is an interdisciplinary field of inquiry that emerges from the need of imbuing autonomous agents with the capacity of moral decision-making. While some approaches provide implementations in Logic Programming (LP) systems, they have not exploited LP-based reasoning features that appear essential for moral reasoning. This PhD thesis aims at investigating further the appropriateness of LP, notably a combination of LP-based reasoning features, including techniques available in LP systems, to machine ethics. Moral facets, as studied in moral philosophy and psychology, that are amenable to computational modeling are identified, and mapped to appropriate LP concepts for representing and reasoning about them. The main contributions of the thesis are twofold. First, novel approaches are proposed for employing tabling in contextual abduction and updating – individually and combined – plus a LP approach of counterfactual reasoning; the latter being implemented on top of the aforementioned combined abduction and updating technique with tabling. They are all important to model various issues of the aforementioned moral facets. Second, a variety of LP-based reasoning features are applied to model the identified moral facets, through moral examples taken off-the-shelf from the morality literature. These applications include: (1) Modeling moral permissibility according to the Doctrines of Double Effect (DDE) and Triple Effect (DTE), demonstrating deontological and utilitarian judgments via integrity constraints (in abduction) and preferences over abductive scenarios; (2) Modeling moral reasoning under uncertainty of actions, via abduction and probabilistic LP; (3) Modeling moral updating (that allows other – possibly overriding – moral rules to be adopted by an agent, on top of those it currently follows) via the integration of tabling in contextual abduction and updating; and (4) Modeling moral permissibility and its justification via counterfactuals, where counterfactuals are used for formulating DDE.