While there is universal agreement that agents ought to act ethically, there is no agreement as to what constitutes ethical behaviour. To address this problem, recent philosophical approaches to `moral uncertainty' propose aggregation of multiple ethical theories to guide agent behaviour. However, one of the foundational proposals for aggregation - Maximising Expected Choiceworthiness (MEC) - has been criticised as being vulnerable to fanaticism; the problem of an ethical theory dominating agent behaviour despite low credence (confidence) in said theory. Fanaticism thus undermines the `democratic' motivation for accommodating multiple ethical perspectives. The problem of fanaticism has not yet been mathematically defined. Representing moral uncertainty as an instance of social welfare aggregation, this paper contributes to the field of moral uncertainty by 1) formalising the problem of fanaticism as a property of social welfare functionals and 2) providing non-fanatical alternatives to MEC, i.e. Highest k-trimmed Mean and Highest Median.

When bias mitigation methods are applied to make fairer machine learning models in fairness-related classification settings, there is an assumption that the disadvantaged group should be better off than if no mitigation method was applied. However, this is a potentially dangerous assumption because a “fair” model outcome does not automatically imply a positive impact for a disadvantaged individual—they could still be negatively impacted. Modeling and accounting for those impacts is key to ensure that mitigated models are not unintentionally harming individuals; we investigate if mitigated models can still negatively impact disadvantaged individuals and what conditions affect those impacts in a loan repayment example. Our results show that most mitigated models negatively impact disadvantaged group members in comparison to the unmitigated models. The domain-dependent impacts of model outcomes should help drive future bias mitigation method development.

In this paper we propose a new model to represent human debates and methods to obtain collective conclusions from them. This model overcomes two drawbacks of existing approaches. First, our model does not assume that participants agree on the structure of the debate. It does this by allowing participants to express their opinion about all aspects of the debate. Second, our model does not assume that participants’ opinions are rational, an assumption that significantly limits current approaches. Instead, we define a weaker notion of rationality that characterises coherent opinions, and we consider different scenarios based on the coherence of individual opinions and the level of consensus. We provide a formal analysis of different opinion aggregation functions that compute a collective decision based on the individual opinions and the debate structure. In particular, we demonstrate that aggregated opinions can be coherent even if there is a lack of consensus and individual opinions are not coherent. We conclude with an empirical evaluation demonstrating that collective opinions can be computed efficiently for real-sized debates.

Language carries implicit human biases, functioning both as a reflection and a perpetuation of stereotypes that people carry with them. Recently, ML-based NLP methods such as word embeddings have been shown to learn such language biases with striking accuracy. This capability of word embeddings has been successfully exploited as a tool to quantify and study human biases. However, previous studies only consider a predefined set of biased concepts to attest (e.g., whether gender is more or less associated with particular jobs), or just discover biased words without helping to understand their meaning at the conceptual level. As such, these approaches can be either unable to find biased concepts that have not been defined in advance, or the biases they find are difficult to interpret and study. This could make existing approaches unsuitable to discover and interpret biases in online communities, as such communities may carry different biases than those in mainstream culture. This paper improves upon, extends, and evaluates our previous data-driven method to automatically discover and help interpret biased concepts encoded in word embeddings. We apply this approach to study the biased concepts present in the language used in online communities and experimentally show the validity and stability of our method.

Smart Personal Assistants (SPAs), such as Amazon Alexa, Google Assistant and Apple Siri, leverage different AI techniques to provide convenient help and assistance to users. However, inappropriate information sharing decisions can lead SPAs to incorrectly disclose user information to undesired parties, or mistakenly block their reasonable access in specific scenarios to desired parties. In fact, reports about privacy violations in SPAs and associated user concerns are well known and understood in the related literature. It is difficult for SPAs to automatically decide how data should be shared with respect to the privacy preferences of the users. We argue norms, which are regarded as shared standards of acceptable behaviour of groups and/or individuals, can be used to govern and reason about the best course of action of SPAs with regards to information sharing, and our work is the first to propose a practical model to address the above issues and govern SPAs based on normative systems and the contextual integrity theory of privacy. We evaluated the performance of the model using a real dataset of user preferences for privacy in SPAs and the results showed a very marked and significant improvement in understanding user preferences and making the right decisions with respect to data sharing.

Norms allow system designers to specify the desired behaviour of a sociotechnical system. In this way, norms regulate what the social and technical agents in a sociotechnical system should (not) do. In this context, a vitally important question is the development of mechanisms for monitoring whether these agents comply with norms. Proposals on norm monitoring often assume that monitoring has no costs and/or that monitors have unlimited resources to observe the environment and the actions performed by agents. In this paper, we challenge this assumption and propose the first practical resource-bounded norm monitor. Our monitor is capable of selecting the resources to be deployed and use them to check norm compliance with incomplete information about the actions performed and the state of the world. We formally demonstrate the correctness and soundness of our norm monitor and study its complexity. We also demonstrate in randomised simulations and benchmark experiments that our monitor can select monitored resources effectively and efficiently, detecting more norm violations and fulfilments than other tractable optimization approaches and obtaining slightly worse results than intractable optimal approaches.

editorial Free Access Share on Guest Editors’ Introduction Authors: Özgür Kafali University of Kent, Canterbury, Kent, UK University of Kent, Canterbury, Kent, UKView Profile , Natalia Criado King’s College London, London, UK King’s College London, London, UKView Profile , Martin Rehak Cisco Systems, Czech Republic Cisco Systems, Czech RepublicView Profile , Jose M. Such King’s College London, London, UK King’s College London, London, UKView Profile , Pinar Yolum Utrecht University, Utrecht, Netherlands Utrecht University, Utrecht, NetherlandsView Profile Authors Info & Claims ACM Transactions on Internet TechnologyVolume 18Issue 3August 2018 Article No.: 26pp 1–4https://doi.org/10.1145/3177884Published:06 March 2018Publication History 0citation462DownloadsMetricsTotal Citations0Total Downloads462Last 12 Months21Last 6 weeks4 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteeReaderPDF

The objective of this study is to explore feature selection for the detection of internal intruders within a local network during the early stages of an attack. As the sophistication of attackers increase, current security systems have proven incapable of detecting advanced stealthy attackers whose aim is to compromise internal networks and remain undetected. We study the available features that are commonly used during network-layer attacker detection and propose two new features to model the extent to which a given networked endpoint conforms with network traffic norms. The proposed features are analysed using several attribute evaluation methods to compare the predictiveness of commonly used features. The results of the analysis show that the proposed features are highly predictive and work towards overcoming the identified deployability issues of previous systems.

In the context of using norms for controlling multiagent systems, a vitally important question that has not yet been addressed in the literature is the development of mechanisms for monitoring norm compliance under partial action observability. This paper proposes the reconstruction of unobserved actions to tackle this problem. In particular, we formalize the problem of reconstructing unobserved actions, and propose an information model and algorithms for monitoring norms under partial action observability using two different processes for reconstructing unobserved actions. Our evaluation shows that reconstructing unobserved actions increases significantly the number of norm violations and fulfilments detected.

The Internet has provided people with new ways of expressing not only their individuality but also their collectivity i.e., their group affiliations. These group identities are the shared sense of belonging to a group. Online contact with others who share the same group identity can lead to cooperation and, even, coordination of social action initiatives both online and offline. Such social actions may be for the purposes of positive change, e.g., the Arab Spring in 2010, or disruptive, e.g., the England Riots in 2011. Stylometry and authorship attribution research has shown that it is possible to distinguish individuals based on their online language. In contrast, this work proposes and evaluates a model to analyse group identities online based on textual conversations amongst groups. We argue that textual features make it possible to automatically distinguish between different group identities and detect whether group identities are salient (i.e., most prominent) in the context of a particular conversation. We show that the salience of group identities can be detected with 95% accuracy and group identities can be distinguished from others with 84% accuracy. We also identify the most relevant features that may enable mal-actors to manipulate the actions of online groups. This has major implications for tools and techniques to drive positive social actions online or safeguard society from disruptive initiatives. At the same time, it poses privacy challenges given the potential ability to persuade or dissuade large groups online to move from rhetoric to action.

Items shared through Social Media may affect more than one user's privacy-e.g., photos that depict multiple users, comments that mention multiple users, events in which multiple users are invited, etc. The lack of multi-party privacy management support in current mainstream Social Media infrastructures makes users unable to appropriately control to whom these items are actually shared or not. Computational mechanisms that are able to merge the privacy preferences of multiple users into a single policy for an item can help solve this problem. However, merging multiple users' privacy preferences is not an easy task, because privacy preferences may conflict, so methods to resolve conflicts are needed. Moreover, these methods need to consider how users' would actually reach an agreement about a solution to the conflict in order to propose solutions that can be acceptable by all of the users affected by the item to be shared. Current approaches are either too demanding or only consider fixed ways of aggregating privacy preferences. In this paper, we propose the first computational mechanism to resolve conflicts for multi-party privacy management in Social Media that is able to adapt to different situations by modelling the concessions that users make to reach a solution to the conflicts. We also present results of a user study in which our proposed mechanism outperformed other existing approaches in terms of how many times each approach matched users' behaviour.

Open Multi-Agent Systems (MAS) are formed by autonomous and heterogeneous agents that may pursue different and even conflicting goals. Therefore, no assumption can be a priori made about their behaviours. Thus, it is needed to constrain the potential excesses of agent autonomous behaviour by means of regulation of agent interaction. Because of this, ‘social’ notions, such as organizations, institutions and norms, have been introduced in the MAS research as coordination and cooperation mechanisms. However, there is still a need to define frameworks which take these organizational abstractions into account. With this aim, this work describes a model of norms for agent organizations that covers both institutional and interaction levels, as well as different types of norms (i.e. constitutive, regulative and procedural norms). This model of norms has been implemented in a specific multi-agent platform, named THOMAS (Methods, Techniques and Tools for Open Multi-Agent Systems), which provides a suitable framework for the management of agent organizations.