A preliminary of this work was presented at HAIDM 2014. The 2014 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

This work was the winner of the “Best Student Paper Award” at the Seventh International Conference on Educational Data Mining (EDM 2014).

Abstract: The growing prevalence of e-learning systems and on-line courses has made educational material widely accessible to students of varying abilities, backgrounds and styles. There is thus a growing need to accommodate for individual differences in such e-learning systems. This paper presents a new algorithm for personalizing educational content to students that combines collaborative filtering algorithms with social choice theory. The algorithm constructs a “difficulty” ranking over questions for a target student by aggregating the ranking of similar students, as measured by different aspects of their performance on common past questions, such as grades, number of retries, and time spent solving questions. It infers a difficulty ranking directly over the questions for a target student, rather than ordering them according to predicted performance, which is prone to error. The algorithm was tested on two large real world data sets containing tens of thousands of students and a million records. Its performance was compared to a variety of personalization methods as well as a non-personalized method that relied on a domain expert. It was able to significantly outperform all of these approaches according to standard information retrieval metrics. Our approach can potentially be used to support teachers in tailoring problem sets and exams to individual students and students in informing them about areas they may need to strengthen..

Citation: EduRank: A Collaborative Filtering Approach to Personalization in E-learning Avi Segal, Ziv Katzir, Ya’akov Gal, Guy Shani and Bracha Shapira. EduRank: A Collaborative Filtering Approach to Personalization in E-learning. Seventh International Conference on Educational Data Mining (EDM 2014), London, England, July 2014.

Download: http://bit.ly/1WzaWlY

This work was presented at HAIDM 2014. The 2014 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: Interaction between a user and a product like a home-appliance is sometimes a mutual relationship. A product not only provides service to a user, but also requires maintenance by the user to keep it working smoothly. If the user stops paying attention to maintenance, the product will no longer be used. To keep a sustainable relationship between a user and a product, the author proposes making a product an agent with anthropomorphic representation and interaction supported by the Anti-max Prisoner’s Dilemma game.

Keywords: Human-agent interaction, human interface, genetic programming.

Citation: Hirotaka Osawa. Sustainable Relationship with Product by Implementing Intentional Interaction.

Download: http://bit.ly/1Snueep

This work was presented at HAIDM 2014. The 2014 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: The multidisciplinary challenge of modelling agents have been driven by theory explaining social phenomena. Yet, these generic models lack of expressiveness. For that reason, data-driven approaches to the design of agents have been pursued, mainly for modelling non-verbal behaviour. In this paper we argue that real data is not only useful for that modality, but it can also assist agent’s design in different phases of the process at different levels of granularity. Furthermore, deep data, which inform us about user’s perception, emotions and motivations is valuable to build fluid interactions with virtual humans. We illustrate our stance with two case studies where we study interpersonal conflict. One study describes the design of agents to populate a serious game aimed at teaching conflict resolution skills to children and the other describes an experiment designed to extract deep data from a dyadic interaction prone to conflict emergence.

Keywords: Virtual Agents, Design, Interpersonal Conflict, Data-driven.

Citation: Joana Campos and Ana Paiva. Interactive Social Agents from Deep Data.

Download: http://bit.ly/1STrWhr

This work was presented at HAIDM 2014. The 2014 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: To date, a variety of automated negotiation agents have been created. While each of these agents has been shown to be effective in negotiating with people in specific environments, they lack natural language processing support required to enable real-world types of interactions. In this paper we present NegoChat, the first negotiation agent that successfully addresses this limitation. NegoChat contains several significant research contributions. First, we found that simply modifying existing agents to include an NLP module is insufficient to create these agents. Instead, the agents’ strategies must be modified to address partial agreements and issue-by-issue interactions. Second, we present NegoChat’s negotiation algorithm. This algorithm is based on bounded rationality, and specifically Aspiration Adaptation Theory (AAT). As per AAT, issues are addressed based on people’s typical urgency, or order of importance. If an agreement cannot be reached based on the value the human partner demands, the agent retreats, or downwardly lowers the value of previously agreed upon issues so that a “good enough” agreement can be reached on all issues. This incremental approach is fundamentally different from all other negotiation agents, including the state-of-the-art KBAgent. Finally, we present a rigorous evaluation of NegoChat, showing its effectiveness.

Keywords: Human-Agent Systems, Negotiation, Chat Agent.

Citation: Avi Rosenfeld. NegoChat: A Chat-Based Negotiation Agent.

Download: http://bit.ly/1XD4IQy

This work was presented at HAIDM 2014. The 2014 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: Recently, many algorithms have been developed for autonomous agents to manage home energy use on behalf of their human owners. By so doing, it is expected that agents will be more efficient at, for example, choosing the best energy tariff to switch to when dynamically priced tariffs come about. However, to date, there has been no validation of such technologies in any field trial. In particular, it has not been shown whether users prefer fully autonomous agents as opposed to controlling their preferences manually. Hence, in this paper we describe a novel platform, called TariffAgent, to study notions of flexible autonomy in the context of tariff switching. TariffAgent uses real-world datasets and real-time electricity monitoring to instantiate a scenario where human participants may have to make, or delegate to their agent (in different ways), tariff switching decisions given uncertainties about their own consumption and tariff prices. We carried out a field trial with 10 participants and, from both quantitative and qualitative results, formulate novel design guidelines for systems that implement flexible autonomy.

Keywords: Human-Agent Interaction, Autonomous Agents, Flexible Autonomy, Energy, Smart Grid.

Citation: Alper Alan, Enrico Costanza, Joel Fisher, Sarvapali Ramchurn, Tom Rodden and Nicholas Jennings. A Field Study of Human-Agent Interaction for Electricity Tariff Switching.

Download: http://bit.ly/1MBoStF

This work was presented at HAIDM 2014. The 2014 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: This paper proposes the Multi-Operation Patrol Scheduling System (MOPSS), a new system to generate patrols for transit system. MOPSS is based on five contributions. First, MOPSS is the first system to use three fundamentally different adversary models for the threats of fare evasion, terrorism and crime, generating three significantly different types of patrol schedule. Second, to handle uncertain interruptions in the execution of patrol schedules, MOPSS uses Markov decision processes (MDPs) in its scheduling. Third, MOPSS is the first system to account for joint activities between multiple resources, by employing the well known SMART security game model that tackles coordination between defender’s resources. Fourth, we are also the first to deploy a new Opportunistic Security Game model, where the adversary, a criminal, makes opportunistic decisions on when and where to commit crimes. Our fifth, and most important, contribution is the evaluation of MOPSS via real-world deployments, providing data from security games in the field.

Keywords: Security, Game-theory, Real-world deployment.

Citation: Francesco Delle Fave, Matthew Brown, Chao Zhang, Eric Shieh, Albert Jiang and Milind Tambe. Security Games in the Field: Deployments on a Transit System.

Download: http://bit.ly/1TcVx85

This work was presented at HAIDM 2014. The 2014 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: This paper studies commitment strategies in three-player negotiation settings comprising human players and computer agents. We defined a new game called the Contract Game which is analogous to real-world market settings in which participants need to reach agreement over contracts in order to succeed. The game comprises three players, two service providers and one customer. The service providers compete to make repeated contract offers to the customer consisting of resource exchanges in the game. We formally analyzed the game and defined sub-game perfect equilibrium strategies for the customer and service providers that involve commitments. We conducted extensive empirical studies of these strategies in three different countries, the U.S., Israel and China. We ran several configurations in which two human participants played a single agent using the equilibrium strategies in various role configurations in the game (both customer and service providers). Our results showed that the computer agent using equilibrium strategies for the customer role was able to outperform people playing the same role in all three countries. In contrast, the computer agent playing the role of the service provider was not able to outperform people. Analysis reveals this difference in performance is due to the contracts proposed in equilibrium being significantly beneficial to the customer players, as well as irrational behavior taken by human customer players in the game.

Citation: Galit Haim, Kobi Gal, Bo An and Sarit Kraus. Equilibrium Strategies for Human-Computer Negotiation in 3-player market settings.

Download: http://bit.ly/1QfXcFa

This work was presented at HAIDM 2014. The 2014 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: This paper deals with platforms that provide agents easier access to the type of opportunities in which they are interested (e.g., eCommerce platforms, used cars bulletins and dating web-sites). We show that under various common service schemes, a platform can benefit from not necessarily listing all the opportunities with which it is familiar, even if there is no marginal cost for listing any additional opportunity. The main implication of this result is that platforms should extract their expected-profit-maximizing service terms not based solely on the fees charged from users, but they should also use the subset that will be listed as the decision variable in the optimization problem. The analysis applies to four well-known service schemes that a platform may use to price its services. We show that neither of these schemes generally dominates the others or is dominated by any of the others. For the common case of homogeneous preferences, however, several dominance relationships can be proved, enabling the platform to identify the schemes that should be used as a default. Furthermore, the analysis provides a game-theoretic search-based explanation for a possible preference of buyers to pay for the service rather than receive it for free (e.g., when the service is sponsored by ads), a phenomena that has been justified in prior literature typically with the argument of willingness to pay a premium for an ad-free experience or more reliable platforms. The paper shows that this preference can hold both for the users and the platform in a given setting, even if both sides are fully strategic.

Keywords: Platforms and Services, Economics of Information, Two-Sided Markets, Price of Free, Service Schemes.

Citation: Chen Hajaj and David Sarne. Strategic Information Platforms – Selective Disclosure and The Price of “Free”.

Download: http://bit.ly/1VwVZSz

This work was presented at HAIDM 2014. The 2014 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: Peer Designed Agent (PDA), computer agents developed by non-experts, is an emerging technology, widely advocated in recent literature for the purpose of replacing people in simulations and investigating human behavior. Its main premise is that the strategy programmed into these agents reliably reflect, to some extent, the behavior used by the programmer in real life. In this paper we show that PDA development has an important side effect that has not been addressed to date — the process, that merely attempts to capture one’s strategy, is also likely to affect the developer’s strategy. The phenomenon is demonstrated experimentally via the penetration detection game, using different setting variations. This result has many implications concerning the appropriate design of PDA-based simulations, and the validness of using PDAs for studying individual decision making.

Keywords: PDAs, decision making, simulation design.

Citation: Avshalom Elmalech, David Sarne and Noa Agmon. Peer Designed Agents: Just reflect or also affect?.

Download: http://bit.ly/1NvNHlq

This work was presented at HAIDM 2014. The 2014 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: Choice selection processes are a family of bilateral games of incomplete information in which a computer agent generates advice for a human user while considering the effect of the advice on the user’s behavior in future interactions. The human and the agent may share certain goals, but are essentially self-interested. This paper extends selection processes to settings in which the actions available to the human are ordered and thus the user may be influenced by the advice even though he doesn’t necessarily follow it exactly. In this work we also consider the case in which the user obtains some observation on the sate of the world. We propose several approaches to model human decision making in such settings. We incorporate these models into two optimization techniques for the agent advice provision strategy. In the first one the agent used a social utility approach which considered the benefits and costs for both agent and person when making suggestions. In the second approach we simplified the human model in order to allow modeling and solving the agent strategy as an MDP. In an empirical evaluation involving human users on AMT, we showed that the social utility approach significantly outperformed the MDP approach.

Keywords: Human modeling, advice provision, persuasion.

Citation: Amos Azaria, Sarit Kraus, Claudia Goldman and Kobi Gal. Advice Provision for Choice Selection Processes with Ranked Options.

Download: http://bit.ly/1U002VV

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: There is increasing interest in the agent community to integrate the concept of emotions and artificial agents. The spectrum of available solutions reaches from applications and models of emotions to complete axiomatised logics. Despite the rich offer of solutions, available works neglect individual personality as a significant factor for the outcome of emotional behaviour pattern. However, different personalities affect all relevant phases of human decision-making processes. Hence, this paper introduces and discusses existing personality theories and highlights the fact that one of them is widely accepted in psychology and should be adopted by the agent-community. We integrate the characteristics of this personality theory into the life-cycle of BDI agents and discuss two different versions of the BDI algorithm – a naive one and one that balances the commitment between means and ends. The outlined algorithm is implemented as a prototype model in AntMe!, an agent-based simulation environment for behavioural studies. The experiments performed in this environment show that personality indeed affects all relevant phases of the decision-making process, laying the foundations for future empirical studies.

Citation: Sebastian Ahrndt, Johannes Fähndrich and Sahin Albayrak. Modelling of Personality in Agents: From Psychology to Implementation.

Download: http://bit.ly/1VzKTfE

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: We study simulation models of spreading on peer-to-peer communication networks where any peer (or agent) can be the source of information, be it sensory recognition or contextual knowledge. In such a situation the value or quality of information is of key relevance. Questions of trust, provenance and the problem of the interaction pattern arise and are approached by three different algorithms in our paper: (i) “quantitative democracy”, where knowledge is averaged on a meeting (ii) “experience takes all”, where the more experienced (the teacher) overwrites all prior knowledge of the less experienced (the “student”), and (iii) “transitive experience” where not only information but also experience is handed over. We compare these different regimes and identify their tradeoffs.

Keywords: Trust, provenance, self-organization, emergence, collaborative information processing.

Citation: George Kampis and Paul Lukowicz. Collaborative Activity Recognition.

Download: http://bit.ly/1SQJhad

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: Trust is an important factor in any relationship and within teamwork is no exception. Teammates need to trust each other to achieve common tasks effectively and efficiently. Teamwork that combines both a human and an Intelligent Virtual Agent (IVA) has drawn much interest; nevertheless, the handling of trust between humans and IVAs is unclear. In this paper, we seek to understand how people trust an IVA teammate. The current study considers two facets of trust: personality and cognition. Our experimental study with 55 participants, involving a collaborative human-IVA task, sought to determine whether human trust in an IVA teammate is affected by the IVA’s personality and whether that differs when the IVA’s personality matches the human’s personality. Furthermore, we sought to understand the relative importance of personality-based versus cognitive-based facets (e.g. the information offered by the IVA) on human trust in the IVA and the resultant effect of human trust on team performance. Results indicated that cognitive-based facets played a more dominant role in establishing trust than personality-based facets. Additionally, the results showed that human trust in the IVA had a significantly positive influence on human-IVA team performance.

Keywords: Intelligent Virtual Agent, Human-Agent Teamwork, Multimodal Communication, Trust, Personality, FFM, Team Performance.

Citation: Nader Hanna and Deborah Richards. In the Agent we Trust! The Role of Personality and Cognition in Human Trust in Virtual Agents.

Download: http://bit.ly/23I9B1H

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: Despite the prevalence of e-learning systems in schools, most of today’s systems do not personalize educational data to the individual needs of each student. This paper proposes a new algorithm for sequencing questions to students that is empirically shown to lead to better performance and engagement in real schools when compared to a baseline approach. It is based on using knowledge tracing to model students’ skill acquisition over time, and to select questions that advance the student’s learning within the range of the student’s capabilities, as determined by the model. The algorithm is based on a Bayesian Knowledge Tracing (BKT) model that incorporates partial credit scores, reasoning about multiple attempts to solve problems, and integrating item difficulty. This model is shown to outperform other BKT models that do not reason about (or reason about some but not all) of these features. The model was incorporated into a sequencing algorithm and deployed in two classes in different schools where it was compared to a baseline sequencing algorithm that was designed by pedagogical experts. In both classes, students using the BKT sequencing approach solved more difficult questions and attributed higher performance than did students who used the expert-based approach. Students were also more engaged using the BKT approach, as determined by their interaction time and number of log-ins to the system, as well as their reported opinion. We expect our approach to inform the design of better methods for sequencing and personalizing educational content to students that will meet their individual learning needs.

Citation: Yossi Ben David, Avi Segal, and Kobi Gal. Sequencing Educational Content in Classrooms using Bayesian Knowledge Tracing.

Download: http://bit.ly/23I8nDB

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: With the improvement in agent technology and agent capabilities we foresee increasing use of agents in social contexts and, in particular, in human-agent team applications. To be effective in such team contexts, agents need to understand and adapt to the expectation of human team members. This paper presents our study on how behavioral strategies of agents affect the humans’ trust in those agents and the concomitant performance expectations that follow in virtual team environments. We have developed a virtual teamwork problem that involves repeated interaction between a human and several agent types over multiple episodes. The domain involves transcribing spoken words, and was chosen so that no specialized knowledge beyond language expertise is required of the human participants. The problem requires humans and agents to independently choose subset of tasks to complete without consulting with the partner and utility obtained is a function of the payment for task, if completed, minus its efforts. We implemented several agents types, which vary in how much of the teamwork they perform over different interactions in an episode. Experiments were conducted with subjects recruited from the MTurk. We collected both teamwork performance data as well as surveys to gauge participants’ trust in their agent partners. We trained a regression model on collected game data to identify distinct behavioral traits. By integrating the prediction model of player’s task choice, a learning agent is constructed and shown to significantly improve both social welfare, by reducing redundant work without sacrificing task completion rate, as well as agent and human utilities.

Keywords: Human-agent interaction, teamwork, trust, adaptation.

Citation: Feyza Hafizoglu and Sandip Sen. Evaluating Trust Levels in Human-agent Teamwork in Virtual Environments.

Download: http://bit.ly/1S8ceCA

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: Interactive agent-based systems are becoming increasingly ubiquitous in our everyday lives, helping people in many domains including healthcare, transportation, and energy. As such, there is a need to investigate how humans and agents interact with each other to maximize the benefit that such systems can offer. In this paper, we present a field study that lasted for six weeks, in which 12 different households were required to interact on a daily basis with an agent-based system in order to manage their electricity pricing scheme. The main goal of this study was to explore long term interactions between human users and an agent to understand how people’s trust towards an agent may change over time, and consequently affect their autonomy and interaction preferences. Our results suggest that flexible autonomy shows promise for sustaining users’ trust and engagement with an agent, despite its occasional mistakes.

Keywords: Human-Agent Interaction, Autonomous Agents, Flexible Autonomy, Energy.

Citation: Alper Alan, Enrico Costanza, Sarvapali Ramchurn, Joel Fischer, Tom Rodden and Nicholas Jennings. Evaluating Human-Agent Interaction in the Wild.

Download: http://bit.ly/1S8bZri

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: Crowdsourcing has emerged as a powerful way to provide computer systems with quick and easy access to human intelligence. However, there is a risk that online crowd workers could be directed to perform harmful tasks. To understand the impact of financial incentives on paid crowd workers’ willingness to behave maliciously, we conducted a series of experiments in which we hired crowd workers via one crowdsourcing task (Attack task) to attack a different crowdsourcing task (Target task. We found that roughly one third of all crowd workers were willing to provide the attack task with potentially sensitive information from the target task, and that we could double this number by increasing the payment of the Attack task. Based on exit interviews and community feedback, we discuss some of what workers reported. Our findings reveal a measurable cost to completing malicious work that well-meaning task designers can leverage to protect their systems from attack.

Citation: Walter Lasecki, Jaime Teevan and Ece Kamar. What’s Your Price? The Cost of Asking Crowd Workers to Behave Maliciously.

Download: http://bit.ly/1VB1DTh

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: Recently, there has been an increase of interest in domains involving repeated interactions between defenders and adversaries. This has been modeled as a repeated Stackelberg Security Game (repeated SSG). Although different behavioral models have been proposed for the attackers in these games, human subjects experiments for testing these behavioral models in repeated SSGs have not been conducted previously. This paper presents the first “longitudinal study” – at least in the context of SSGs – of testing human behavior models in repeated SSG settings. We provide the following contributions in this paper. First, in order to test the behavioral models, we design a game that simulates the repeated interactions between the defender and the adversary and deploy it on Amazon Mechanical Turk (AMT). Human subjects are asked to participate in this repeated task in rounds of the game, with a break between consecutive rounds. Second, we develop several approaches to keep the human subjects motivated throughout the course of this longitudinal study so that they participate in all measurement occasions, thereby minimizing attrition. We provide results showing improvements of retention rate due to implementation of these approaches. Third, we propose a way of choosing representative payoffs that fit the real-world scenarios as conducting these experiments are extremely time-consuming and we can only conduct a limited number of such experiments.

Keywords: Game Theory, Human Behavior Models, Repeated Stackelberg Games, Longitudinal Experiments, Amazon Mechanical Turk.

Citation: Debarun Kar, Fei Fang, Francesco Delle Fave, Nicole Sintov and Milind Tambe. Conducting Longitudinal Experiments with Behavioral Models in Repeated Stackelberg Security Games on Amazon Mechanical Turk.

Download: http://bit.ly/1rf3wsu

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: Humans use sketches drawn on paper, on a computer, or via hand gestures in the air as part of their communications. To recognize shapes in sketches, most existing work focuses on offline (post-drawing) recognition methods, trained on large sets of examples which serve as a plan library for the recognition method. These methods do not allow on-line recognition, and require a very large library (or expensive pre-processing) in order to recognize shapes that have been translated, rotated or scaled. Inspired by mirroring processes in human brains we present an online shape recognizer that identifies multi-stroke geometric shapes without a plan library. Instead, the recognizer uses a shape-drawing planner for drawn-shape recognition, i.e., a form of plan recognition by planning. This method (1) allows recognition of shapes that is immune to geometric translations, rotations, and scale; (2) eliminates the need for storing a library of shapes to be matched against drawings (instead, only needs a set of possible Goals and a planner that can instantiate them in any manner); and (3) allows fast on-line recognition. The method is particularly suited to complete agents, that must not only recognize sketches, but also produce them, and therefore necessarily have a drawing planner already. We compare the performance of different variants of the recognizer to that of humans, and show that its recognition level is close to that of humans, while making less recognition errors early in the recognition process.

Citation: Mor Vered and Gal Kaminka. If You Can Draw It, You Can Recognize It: Mirroring For Sketch Recognition.

Download: http://bit.ly/1WcL6UC

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: Social interaction has essential effects on social reality, which can even outweigh the importance of the physical effects, but which are not directly nor objectively observable. Understanding the social contexts in which actions and interactions take place is thus of utmost importance for planning one’s goals and activities. We claim that joint action should be considered from both the social and the physical perspective jointly. These aspects are interdependent and influence each other continuously. In order to support the inclusion of social aspects into state descriptions for joint actions, we propose a methodological approach for social analysis of joint action, that enables to identify and represent the social characteristics of a joint action setting. Through the use of social practices it is possible to combine both physical and social aspects of joint actions. We show how these social practices can be used to design agents and robots that take into account both the social and physical context and goals.

Citation: Virginia Dignum, Frank Dignum and Catholijn Jonker. Social Aspects of Joint Actions: from analysis to design of social actions.

Download: http://bit.ly/22KPwCa

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: It can be hard to automatically identify sensitive content in images or other media because significant context is often necessary to interpret noisy content and complex notions of sensitivity. Online crowds can help computers interpret information that cannot be understood algorithmically. However, systems that use this approach can unwittingly show workers information that should remain private. For instance, images sent to the crowd may accidentally include faces or geographic identifiers in the background, and information pertaining to a task (e.g., the amount of a bill) may appear alongside private information (e.g., an account number). This paper introduces an approach for using crowds to filter information from sensory data that should remain private, while retaining information needed to complete a specified task. The pyramid workflow that we introduce allows crowd workers to identify private information while never having complete access to the (potentially private) information they are filtering. Our approach is flexible, easily configurable, and can protect user information in settings where automated approaches fail. Our experiments with 4685 crowd workers show that it performs significantly better than previous approaches.

Citation: Walter Lasecki, Mitchell Gordon, Jaime Teevan, Ece Kamar and Jeffrey Bigham. CrowdMask: Privacy-Preserving Crowd-Powered Systems.

Download: http://bit.ly/1YF37Ke

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: While the development of techniques that allow artificial agents to engage in dialogue with humans has received a lot of interest in the multiagent systems and natural language processing literature, most of the systems created to date have focused on specific domains and types of dialogue. This has led to agent designs that are useful for specific dialogue situations, but hard to adapt to different settings. The creation of more flexible agents that can deal with a broad range of communicative scenarios would greatly improve the interaction between agents and humans, and would eliminate the need to manually adapt the design of a conversational agent when dealing with a new task domain. In this paper, we present initial work toward creating agents that are able to generate task-oriented dialogues based on a description of a previously unknown domain. Our method is based on utilising automated planning methods, which are suitable for processing specifications both of the communication language to be used and of the domain in hand. We provide preliminary experimental results which suggest that our method has the potential to provide the flexibility required to produce a broad range of communication behaviours in different settings.

Keywords: Dialogue Planning, Conversational Agents, Human-Agent Dialogue.

Citation: Tânia Marques and Michael Rovatsos. Toward Domain-Independent Dialogue Planning.

Download: http://bit.ly/1VzKPg1

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: The majority of volunteers participating in citizen science projects perform only a few tasks each before leaving the system. We designed an intervention strategy to reduce disengagement in 16 different citizen science projects. Targeted users who had left the system received emails that directly addressed motivational factors that affect their engagement. Results show that participants receiving the emails were significantly more likely to return to productive activity when compared to a control group.

Keywords: Peer production, crowdsourcing, citizen science, intervention strategies.

Citation: Segal, A., Gal, Y.A.K., Simpson, R.J., Victoria Homsy, V., Hartswood, M., Page, K.R. and Jirotka, M., 2015, May. Improving productivity in citizen science through controlled intervention. In Proceedings of the 24th International Conference on World Wide Web Companion (pp. 331-337). International World Wide Web Conferences Steering Committee.

Download: http://bit.ly/1SiqBku

This work was presented at HAIDM 2015. The 2015 workshop on Human-Agent Interaction Design and Models was co-organised by SmartSociety.

Abstract: Preventive maintenance is essential for the smooth operation of any equipment. Still, people occasionally do not maintain their equipment adequately. Alert systems attempt to remind people to perform maintenance. However, most of these systems do not provide alerts at the optimal timing, and nor do they take into account the time required for maintenance or compute the optimal timing for a specific user. In this paper we model the problem of maintenance performance, assuming maintenance is time consuming. We solve the optimal policy for the user, i.e., the optimal timing for a user to perform maintenance. This optimal strategy depends on the user’s value of time, and thus it may vary from user to user and may change over time. Based on the solved optimal strategy we present a personalized alert agent, which, depending on the user’s value of time, alerts the user when she should perform maintenance. In an experiment using a spaceship computer game, we show that receiving alerts from the personalized alert agent significantly improves user performance.

Citation: Avraham Shvartzon, Amos Azaria, Sarit Kraus, Claudia Goldman, Joachim Meyer and Omer Tsimhoni. Worth Fixing: Personalizing Maintenance Alerts for Optimal Performance.

Download: http://bit.ly/1WcKUVa