Formation conditions of mutual adaptation in human-agent collaborative interaction

When an adaptive agent works with a human user in a collaborative task, in order to enable flexible instructions to be issued by ordinary people, it is believed that a mutual adaptation phenomenon can enable the agent to handle flexible mapping relations between the human user’s instructions and the agent’s actions. To elucidate the conditions required to induce the mutual adaptation phenomenon, we designed an appropriate experimental environment called “WAITER” (Waiter Agent Interactive Training Experimental Restaurant) and conducted two experiments in this environment. The experimental results suggest that the proposed conditions can induce the mutual adaptation phenomenon.     

A data-driven passing interaction model for embodied basketball agents

Human beings have an ability to transition smoothly between individual and collaborative activities and to recognize these types of activity in other humans. Our long-term goal is to devise an agent which can function intelligently in an environment with frequent switching between individual and collaborative tasks. A basketball scenario is such an environment, however there currently do not exist suitable interactive agents for this domain. In this paper we take a step towards intelligent basketball agents by contributing a data-driven generalized model of passing interactions. We first collect data on human-human interaction in virtual basketball to discover patterns of behavior surrounding passing interactions. Through these patterns we produce a model of rotation behavior before and after passes are executed. We then implement this model into an actual basketball agent and then conduct an experiment with a human-agent team. Results show that the agent using the model can at least communicate better than a task-competent agent with limited communication, with participants rating the agent as being able to recognize and express its intention. In addition we analyze passing interactions using Herbert Clark’s joint activity theory and propose that the concepts, while completely theoretical, should be considered as a basis for agent design.     

Building intelligent collaborative interface agents with the ICAGENT development framework

This paper describes the implementation of intelligent collaborative interface agents using the intelligent collaborative agent (ICagent) development framework. In particular, the paper presents the implementation of a collaborative interface agent that acts as a tutor in the context of an educational software application. The agent deliberates socially with users following the SharedPlans model of collaborative activity. Social deliberation requires interface agents to make their desires and intentions clear to the application users, being in constant communication with them, to understand the context of their activity and to reconcile their own and users’ desires in the overall context of action. Reconciliation of users’ desires allows agents to recognize the situations where users need help. The paper briefly presents the ICagent development framework, describes the implementation of the interface agent, and discusses an example of the behavior of the agent during a collaboration session.     

Close Encounters of the Virtual Kind: Agents Simulating Copresence

ABSTRACT

Results of a field study of an open-access collaborative virtual environment in actual use suggested that awareness of others significantly increases the level of presence experienced by participants. Given the importance of copresence, this paper argues that, in the absence of other human collaborators in a collaborative virtual environment, copresence can potentially be simulated using agent technology. A controlled experiment deploying a prototype embodied conversational agent was conducted to investigate the potential of such agents to simulate copresence. This paper briefly introduces the concepts of presence and copresence, summarizes experiences drawn from the field study, reports on the controlled experiment, and discusses its results. Results suggest that even limited copresence as provided by the current prototype agent is sufficient to help users feel presence in the environment.     

NDM-Based Cognitive Agents for Supporting Decision-Making Teams

Naturalistic decision making (NDM) focuses on how people actually make decisions in realistic settings that typically involve ill-structured problems. Taking an experimental approach, we investigate the impacts of using an NDM-based software agent (R-CAST) on the performance of human decision-making teams in a simulated C3I (Communications, Command, Control and Intelligence) environment. We examined four types of decision-making teams with mixed human and agent members playing the roles of intelligence collection and command selection. The experiment also involved two within-group control variables: task complexity and context switching frequency. The result indicates that the use of an R-CAST agent in intelligence collection allows its team member to consider the latest situational information in decision making but might increase the team member's cognitive load. It also indicates that a human member playing the role of command selection should not rely too much on the agent serving as his or her decision aid. Together, it is suggested that the roles of both humans and cognitive agents are critical for achieving the best possible performance of C3I decision-making teams: Whereas agents are superior in computation-intensive activities such as information seeking and filtering, humans are superior in projecting and reasoning about dynamic situations and more adaptable to teammates' cognitive capacities. This study has demonstrated that cognitive agents empowered with NDM models can serve as the teammates and decision aids of human decision makers. Advanced decision support systems built upon such team-aware agents could help achieve reduced cognitive load and effective human-agent collaboration.     

Effective agent collaboration through improved communication by means of contextual reasoning

This paper describes an extension to a context‐driven agent representation paradigm that facilitates modeling collaborative tactical behaviors for simulations of team games or military missions. Called collaborative context‐based reasoning, it emphasizes communication among the collaborating agents and carries it out by exchanging their currently active context when feasible. CCxBR is founded on the concepts defined in joint intention theory (JIT). The research described here presents an architecture that incorporates JIT in a contextual framework. The ability to facilitate communication among the collaborating agents by exchanging information about active contexts resembles the ability of humans to agree on a tactic in midstream and predict the behavior of their collaborators. This allows a CCxBR agent to invoke the actions involved in the tactic in the pursuit of a common goal. The paper describes several prototypes built to evaluate the CCxBR approach and the experiments executed to determine its effectiveness. The results of the experiments and the conclusions reached are discussed. © 2010 Wiley Periodicals, Inc.     

Experiments Toward a Mutual Adaptive Speech Interface That Adopts the Cognitive Features Humans Use for Communication and Induces and Exploits Users' Adaptations

Interactive agents such as pet robots or adaptive speech interface systems that require forming a mutual adaptation process with users should have two competences. One of these is recognizing reward information from users' expressed paralanguage information, and the other is informing the learning system about the users by means of that reward information. The purpose of this study was to clarify the specific contents of reward information and the actual mechanism of a learning system by observing how 2 persons could create a smooth speech communication, such as that between owners and their pets.

A communication experiment was conducted to observe how human participants create smooth communication through acquiring meaning from utterances in languages they did not understand. Then, based on experimental results, a meaning-acquisition model that considers the following 2 assumptions was constructed: (a) To achieve a mutual adaptive relationship with users, the model needs to induce users' adaptation and to exploit this induced adaptation to recognize the meanings of a user's speech sounds; and (b) to recognize users' utterances through trial-and-error interaction regardless of the language used, the model should focus on prosodic information in speech sounds, rather than on the phoneme information on which most past interface studies have focused.

The results confirmed that the proposed model could recognize the meanings of users' verbal commands by using participants' adaptations to the model for its meaning-acquisition process. However, this phenomenon was observed only when an experimenter gave the participants appropriate instructions equivalent to catchphrases that helped users learn how to use and interact intuitively with the model. Thus, this suggested the need for a subsequent study to discover how to induce the participants' adaptations or natural behaviors without giving these kinds of instructions.     

Are interface agents scapegoats? Attributions of responsibility in human–agent interaction

This paper presents an investigation of the self-serving biases of interface agent users. An experiment that involved 202 MS Office users demonstrated that, in contrast to the self-serving hypothesis in attribution theory, people do not always attribute the successful outcomes of human–agent interaction to themselves and negative results to interface agents. At the same time, it was found that as the degree of autonomy of MS Office interface agents increases, users tend to assign more negative attributions to agents under the condition of failure and more positive attributions under the condition of success. Overall, this research attempts to understand the behavior of interface agent users and presents several conclusions that may be of interest to human–computer interaction researchers and software designers working on the incorporation of interface agents in end-user systems.     

Designing a goal-oriented smart-home environment

Nowadays, systems are growing in power and in access to more resources and services. This situation makes it necessary to provide user-centered systems that act as intelligent assistants. These systems should be able to interact in a natural way with human users and the environment and also be able to take into account user goals and environment information and changes. In this paper, we present an architecture for the design and development of a goal-oriented, self-adaptive, smart-home environment. With this architecture, users are able to interact with the system by expressing their goals which are translated into a set of agent actions in a way that is transparent to the user. This is especially appropriate for environments where ambient intelligence and automatic control are integrated for the user’s welfare. In order to validate this proposal, we designed a prototype based on the proposed architecture for smart-home scenarios. We also performed a set of experiments that shows how the proposed architecture for human-agent interaction increases the number and quality of user goals achieved.     

Integration of Adaptable and Adaptive Approaches for Interface Personalization Through Collaborative Menu

This article examines different user-system collaboration models in the adaptation of a menu interface. Four collaboration models were implemented on a prototype of mobile phone menu: (a) basic collaboration with no system support (for user adaptation) and no user control (over system adaptation), (b) system support only, (c) user control only, and (d) system support plus user control. The prototype mobile phone menu includes a hotlist (a quickly accessible collection of menu items) as well as a hierarchical menu. The hotlist is collaborative, because it combines adaptable and adaptive approaches by allowing both the user and the system to manage the items in it. A controlled experiment compared different types of collaborative menus in order to investigate the effects of system support and user control. Twenty participants performed menu selection tasks in the experiment, and both performance and subjective measures were taken. The results showed that, in a certain condition, the system support and the user control improved the user performance when applied independently, but their effects were not additive. Although the effects disappeared when the selection frequency distribution changed, the system support was preferred by most of the users. The advantages and disadvantages of the collaborative menus and implications for the adaptation of menus are discussed.     

Using hybrid scheduling for the semi-autonomous formation of expert teams

Selecting and scheduling human experts to cooperatively solve a problem can be a highly complex task, given various constraints (such as what expertise is needed and when) and preferences (such as which expertise an expert most prefers to exercise). Computational agents can thus greatly help users form and schedule expert teams. This paper introduces a new formulation of the team formation and scheduling problem as a Hybrid Scheduling Problem (HSP) and compares the performance of an agent using the HSP formulation to a prior agent-based approach. We empirically demonstrate the promise of the HSP formulation and highlight how the application of HSP techniques to this problem has led us to identify important modifications to mechanisms that improve HSP solving. Finally, we summarize how the HSP formulation can support human-agent collaboration during the process of forming and scheduling expert teams.     

Social modification using implementation of partial agency toward objects

This article considers what kind of partial agency can be implemented for objects to bring about better agencies for interacting with humans. We humans have the ability to inform our fellows about our intentions, internal states, and requirements through verbal means, gestures, attitudes, timings, and other representations. These representations help us to maintain our belief that we are sufficient agents. Robots and virtual agents also mimic these representations; they act as if they have such an agency. However, their agencies are sometimes too excessive compared to their task. This mismatch leads to a high cognitive load being placed on users and consequently leads to breakdowns in interaction; it prevents human-agent interaction from being a modality in certain applications. We have devised an agency with multiple selectable features. We believe that selectable features promote good designs of virtual agents, robots, machinery, and home appliances according to their intended traits. We categorized these agencies into several groups and discuss what elements lead to these features. The article also describes a method of identifying these features in human behavior.     

Defining new markets for intelligent agents

Most agent applications are fairly straightforward: access a Web site, fetch material; in short, perform a simple fixed mission. Others do more personalized tasks such as filtering e-mail or updating legacy systems. From a programming viewpoint, agents are simply active objects that have been defined to simulate parts of a model. Agent based modeling and simulation then become a natural extension of the object oriented paradigm. Simulations of events that involve these kinds of agents (known as actors or demons) have assisted human decision making for decades in batch manufacturing, transportation, and logistics, for example. But work in complex adaptive systems (CAS) may be defining a new kind of agent: one that can actually evolve over time in response to its environment. The beginnings of these adaptive systems are already evident in more advanced agents, which can do simple negotiations on a user's behalf to secure goods and services in an auction, for example. The challenge now is to see how agents bargain and learn in a more complex environment. The Electric Power Research Institute (EPRI), for example, has funded research into agent based auctioning as a way to address the fierce competition for resources. As electric power marketers become available, wholesale electric customers are learning to shop around for the best suppliers. Like agents that represent individual human users, the agents bargaining on behalf of these suppliers and wholesalers decide things like how much to buy, which agent to buy from, how much to pay, and how to manage the exchange of power and money     

Complementary menus: Combining adaptable and adaptive approaches for menu interface

This study proposes a method of coupling adaptable and adaptive approaches to the design of menus. The proposed complementary menu types incorporate both adaptability and adaptivity by dividing and allocating menu adaptation roles to the user and the system. Four different types of interface adaptation (i.e., adaptable with/without system support and adaptive with/without user control) were defined. They were implemented in a hypothetical prototype mobile phone via a hotlist (an additional collection of quickly accessible items). A controlled lab experiment was conducted to compare the menu types and investigate the effects of the system support in the adaptable menus and the user control in the adaptive menus. Twenty subjects participated in the experiment and performed menu selection tasks. Both performance and user satisfaction measures were collected. The results showed that adaptable and adaptive menus were superior to the traditional one in terms of both performance and user satisfaction. Providing system support to the adaptable menu not only increased the users’ perception of the efficiency of selection, but also reduced the menu adaptation time. Important implications for the design of menus are described and valuable insights into the menu interface adaptation were gained from the quantitative and qualitative analyses of the experimental results.

Relevance to industry

The evaluation experiment conducted in this study may provide valuable information to designers of adaptive or adaptable menus. Adding system support to adaptable menu would be an attractive option to consider. Also, the results of a user survey provide useful information to the practitioners in mobile phone industry on the features users accessed most frequently.     

An evolutionary approach to prototyping pedagogical agents: from simulation to integrated system

We have developed and integrated software agents with two educational groupware systems (TeamWave Workplace and FLE), using evolutionary prototyping and empirical-based design as development techniques. The resulting prototypes of pedagogical agents (CoPAS, SA-Agent and RuleEditor) provide learners and teachers with increasingly domain-specific support for distributed collaborative learning activities. Employing the evolutionary approach has enabled us to build and evaluate early prototypes of complex systems with cost-effective techniques and involving users in this process helped us to constrain the design space and direct further development. CoPAS is a simulation experiment carried out with the Wizard of Oz technique, SA-Agent is a pedagogical agent integrated with an open-source learning environment, and RuleEditor is a customizer for the SA-Agent. The agents collect statistical information on user activity and analyze that information based on principles of collaboration and knowledge building. The results are presented as advice in the user interface of the learning environments to promote students' reflection on their collaboration and knowledge-building activities. If instructors disagree about the phrasing of the advice or the frequency of intervention, they can change it using the RuleEditor agent customizer.     

Designing E-Collaboration Technologies to Facilitate Compensatory Adaptation

This article argues that e-collaboration technologies often pose obstacles to effective communication in complex collaborative tasks. The reason presented is that typically those technologies selectively suppress face-to-face communication elements that human beings have been designed by evolution to use extensively while communicating with each other. It is argued that technology users invariably react to those obstacles by engaging in compensatory adaptation, whereby they change their communicative behavior in order to compensate for the obstacles. The article concludes with a call for more research on how e-collaboration technologies can be designed to facilitate compensatory adaptation.     

Responsible agent behavior: a distributed computing perspective

Software agents that are autonomous, communicative, and possibly intelligent processes raise new questions for developers of distributed systems. Specifically, what is responsible agent behavior, and who, as the owner, is legally responsible for it? The answers involve an understanding of human-agent interaction, agent-oriented middleware, and social behavior. Some software agents will have a sufficiently large number of internal states to be capable of seemingly intelligent behavior. Hence, an agent's future external behavior cannot be guaranteed on the basis of its past behavior, even if that behavior has been monitored over time. Complete compliance tests of intelligent agents, therefore, may not be achievable because of the (possibly) large number of internal states. Thus, the best we can say is that an agent has not exhibited noncompliant behavior yet. Communication between agents implies a contract between owners, and the complexity of agents implies possibly unpredictable behavior. Therefore, an appropriate legal framework is required to underwrite the consequences of communicative actions and to provide safeguards against unlawful activities. The legal implications of agent technology require new ways of thinking about working with an agent, new requirements for agent-oriented middleware, and additional types of social behavior to be considered when designing a multiagent system     

A note of caution regarding anthropomorphism in HCI agents

Universal usability is an important component of HCI, particularly as companies promote their products in increasingly global markets to users with diverse cultural backgrounds. Successful anthropomorphic agents must have appropriate computer etiquette and nonverbal communication patterns. Because there are differences in etiquette, tone, formality, and colloquialisms across different user populations, it is unlikely that a generic anthropomorphic agent would be universally appealing. Additionally, because anthropomorphic characters are depicted as capable of human reasoning and possessing human motivations, users may ascribe undue trust in these agents. Trust is a complex construct that exerts an important role in a user’s interactions with an interface or system. Feelings and perceptions about an anthropomorphic agent may impact the construction of a mental model about a system, which may lead to inappropriate calibrations of automation trust that is based on an emotional connection with the anthropomorphic agent rather than on actual system performance.     

Adaptive systems: from intelligent tutoring to autonomous agents

Computer systems which can automatically alter aspects of their functionality or interface to suit the needs of individuals or groups of users have appeared over the years in a variety of guises. Most recently, attention has focused on intelligent interface agents, which are seen as specialised, knowledge-based systems acting on behalf of the user in some aspect of the interaction. Similar requirements for automatic adaptation have been noted in intelligent tutoring systems, natural-language systems and intelligent interfaces. The paper brings together the research which has emanated from a number of backgrounds, and provides a unifying perspective on adaptive systems in general. An architecture for adaptive systems and a methodology for their development are presented. The paper also describes software support for producing adaptive systems, and offers some experimental evidence to justify both the desirability and feasibility of exploiting an adaptive system approach to human-computer interaction     

A hybrid collaborative filtering method for multiple-interests and multiple-content recommendation in E-Commerce

Recommender systems apply knowledge discovery techniques to the problem of making personalized recommendations for products or services during a live interaction. These systems, especially collaborative filtering based on user, are achieving widespread success on the Web. The tremendous growth in the amount of available information and the kinds of commodity to Web sites in recent years poses some key challenges for recommender systems. One of these challenges is ability of recommender systems to be adaptive to environment where users have many completely different interests or items have completely different content (We called it as Multiple interests and Multiple-content problem). Unfortunately, the traditional collaborative filtering systems can not make accurate recommendation for the two cases because the predicted item for active user is not consist with the common interests of his neighbor users. To address this issue we have explored a hybrid collaborative filtering method, collaborative filtering based on item and user techniques, by combining collaborative filtering based on item and collaborative filtering based on user together. Collaborative filtering based on item and user analyze the user-item matrix to identify similarity of target item to other items, generate similar items of target item, and determine neighbor users of active user for target item according to similarity of other users to active user based on similar items of target item.In this paper we firstly analyze limitation of collaborative filtering based on user and collaborative filtering based on item algorithms respectively and emphatically make explain why collaborative filtering based on user is not adaptive to Multiple-interests and Multiple-content recommendation. Based on analysis, we present collaborative filtering based on item and user for Multiple-interests and Multiple-content recommendation. Finally, we experimentally evaluate the results and compare them with collaborative filtering based on user and collaborative filtering based on item, respectively. The experiments suggest that collaborative filtering based on item and user provide better recommendation quality than collaborative filtering based on user and collaborative filtering based on item dramatically.