Search

Recommender systems are now popular both commercially and in the research community, where many approaches have been suggested for providing recommendations. In many cases a system designer that wishes to employ a recommendater system must choose between a set of candidate approaches. A first step towards selecting an appropriate algorithm is to decide which properties of the application to focus upon when making this choice. Indeed, recommender systems have a variety of properties that may affect user experience, such as accuracy, robustness, scalability, and so forth. In this paper we discuss how to compare recommenders based on a set of properties that are relevant for the application. We focus on comparative studies, where a few algorithms are compared using some evaluation metric, rather than absolute benchmarking of algorithms. We describe experimental settings appropriate for making choices between algorithms. We review three types of experiments, starting with an offline setting, where recommendation approaches are compared without user interaction, then reviewing user studies, where a small group of subjects experiment with the system and report on the experience, and finally describe large scale online experiments, where real user populations interact with the system. In each of these cases we describe types of questions that can be answered, and suggest protocols for experimentation. We also discuss how to draw trustworthy conclusions from the conducted experiments. We then review a large set of properties, and explain how to evaluate systems given relevant properties. We also survey a large set of evaluation metrics in the context of the property that they evaluate.

User modeling (UM) has traditionally been concerned with analyzing a user’s interaction with a system and with developing cognitive models that aid in the design of user interfaces and interaction mechanisms. Elements of a user model may include representation of goals, plans, preferences, tasks, and/or abilities about one or more types of users, classification of a user into subgroups or stereotypes, the formation of assumptions about the user based on the interaction history, and the generalization of the interaction histories of many users into groups, among many others.

Speech-activated self-service systems in corporate call centers can ­provide callers with an experience that is much closer to the ideal of talking with an experienced agent than is possible with TouchTone™ systems. The closer analogy to live conversation has its benefits, but can also introduce pitfalls for the novice designer. In this chapter, we look at the expectations that callers bring to these phone calls, ranging from broad expectations with regard to self-service in general, to the more specific expectations of human-to-human conversation about ­consumer issues. We recommend several steps to the system designer to produce more successful interaction between callers and speech interfaces. They focus on the thoughtful use of user modeling achieved by employing ideas and concepts related to transparency, choice, and expert advice, all of which most, if not all, callers are already familiar with from their own everyday experiences.

Rich Internet Applications significantly raise the user experience compared with legacy page-based Web applications because of their highly responsive user interfaces. Although this is a tremendous advance, it does not solve the problem of the one-size-fits-all approach1 of current Web applications. So although Rich Internet Applications put the user in a position to interact seamlessly with the Web application, they do not adapt to the context in which the user is currently working. In this paper we address the on-the-fly personalization of Rich Internet Applications. We introduce the concept of ARRIAs: Adaptive Reactive Rich Internet Applications and elaborate on how they are able to adapt to the current working context the user is engaged in. An architecture for the ad hoc adaptation of Rich Internet Applications is presented as well as a holistic framework and tools for the realization of our on-the-fly personalization approach. We divided both the architecture and the framework into two levels: offline/design-time and online/run-time. For design-time we explain how to use ontologies in order to annotate Rich Internet Applications and how to use these annotations for conceptual Web usage mining. Furthermore, we describe how to create client-side executable rules from the semantic data mining results. We present our declarative lightweight rule language tailored to the needs of being executed directly on the client. Because of the event-driven nature of the user interfaces of Rich Internet Applications, we designed a lightweight rule language based on the event–condition–action paradigm.2 At run-time the interactions of a user are tracked directly on the client and in real-time a user model is built up. The user model then acts as input to and is evaluated by our client-side complex event processing and rule engine.

As our ability to access the abundance of clinical data grows, it is imperative that methods to organize and to visualize this information be in place so as not to overwhelm users: increasingly, users are faced with information overload. Moreover, the manner of presentation is fundamental to how such information is interpreted, and can be the turning point in uncovering new insights and knowledge about a patient or a disease. And of course, medical imaging is itself an inherently visual medium. This chapter presents work related to the visualization of medical data, focusing on issues related to navigation and presentation by drawing upon imaging and other disciplines for examples of display and integration methods. We first cover different visual paradigms that have been developed (e.g., icons, graphs), grouped along dimensions that emphasize the different types of data relationships and workflow. Subsequently, issues related to combining these visualizations are given. As no single graphical user interface (GUI) can accommodate all users and the spectrum of tasks seen in the healthcare environment, the ultimate goal is to create an adaptive graphical interface that integrates clinical information so as to be conducive to a given user's objectives: efforts in this direction are discussed. Throughout, we describe applications that illustrate the many open issues revolving around medical data visualization.

In this paper we try to identify requirements, opportunities and problems in home media centers and we propose an approach to address them by describing an intelligent home media environment. The major issues investigated are coping with the information overflow in the current provision of TV programs and channels and the need for personalization to specific users by adapting to their age, interests, language abilities, and various context characteristics. The research presented in this paper follows from a collaboration between Eindhoven University of Technology, the Philips Applied Technologies group and Stoneroos Interactive Television. The work has been partially carried out within the ITEA-funded European project Passepartout, which also includes partners like Thomson, INRIA and ETRI. In the following chapter we describe the motivation and research problem in relation to related work, followed by an illustrative use case scenario. Afterwards, we explain our data model which starts with explaining the TV-Anytime structure and its enrichments with semantic knowledge from various ontologies and vocabularies. The data model description then serves as the background for understanding our proposed system architecture SenSee. Afterwards we go deeper into the user modeling part and explain how our personalization approach works. The latter elaborates on a design targeting interoperability and on semantic techniques for enabling intelligent context-aware personalization. In the implementation chapter we describe some practical issues as well as our main interface showcase, iFanzy. Future work and conclusions end this chapter.

A pervasive computing environment typically comprises multiple embedded devices that may interact together and with mobile users. These users are part of the environment, and they experience it through a variety of devices embedded in the environment. This perception involves technologies which may be heterogeneous, pervasive, and dynamic. Due to the highly dynamic properties of such environments, the software systems running on them have to face problems such as user mobility, service failures, or resource and goal changes which may happen in an unpredictable manner. To cope with these problems, such systems must be autonomous and self-managed. In this chapter we deal with a special kind of a ubiquitous environment, a smart home environment, and introduce a user-preferencebased model for adaptation planning. The model, which dynamically forms a set of configuration plans for resources, reasons automatically and autonomously, based on utility functions, on which plan is likely to best achieve the user's goals with respect to resource availability and user needs. © 2009 Springer Science+Business Media, LLC. 529 537 Conference code: 86223 Cited By :2

In this paper, we propose a novel site customization model that relies on a topical ontology in order to learn the user interests as these are exemplified in their site navigations. Based on this knowledge, our mechanism personalizes the site’s content and structure so as to meet particular user needs. Experimental results demonstrate that our model has a significant potential in accurately identifying the user interests and show that site customizations that rely on the detected interests assist web users experience personalized navigations in the sites’ contents.

During the combination of knowledge and capital, science and technology intermediaries are demanded to provide information integration services. Science and technology intermediaries play an important role in quickening the commercialization of research findings, developing new and high technology industry, and promoting regional scientific and technological innovations. With the influences of the computer age on science and technology information services as the starting point, the thesis makes a large-scale survey of websites of state and provincial comprehensive science and technology intermediaries, and based on this, the overall framework of scientific and technological information integration services under the E-environment is established, and furthermore, the functions and structures of different subsystems are systemically analyzed in the modes of scientific and technological information integration services.

What are the key success factors for a knowledge management system (KMS), and how to design and implement successful knowledge management systems, are topical research areas. We argue that designing effective knowledge management systems requires not only a focused view, which is achieved by considering organizational imperatives and technological solutions, but it also benefits from a larger perspective that considers a user-centered design, the individual needs of the users (e.g. work tasks, responsibilities), individual motivational drivers, usability and ergonomics issues. This article emphasizes the role of user models and user modeling within Ontology-based Knowledge Management System (OKMS), integrating a highly interdisciplinary approach. It shows how user models, models of the knowledge workers and user modeling processes can be applied in the context of knowledge management systems. An ontology-based user modeling approach is proposed and concrete examples of how ontology-based inferences can be used for expertise modeling are provided. This chapter emphasizes the importance of using ontology-based representations for modeling the users and providing enhanced user support and advanced features in KMSs.

The ParetoGP algorithm which adopts a multi-objective optimization approach to balancing expression complexity and accuracy has proven to have significant impact on symbolic regression of industrial data due to its improvement in speed and quality of model development as well as user model selection, (Smits and Kotanchek, 2004), (Smits et al., 2005), (Castillo et al., 2006). In this chapter, we explore a range of topics related to exploiting the Pareto paradigm. First we describe and explore the strengths and weaknesses of the ClassicGPand Pareto-Front GP variants for symbolic regression as well as touch on related algorithms. Next, we show a derivation for the selection intensity of tournament selection with multiple winners (albeit, in a single-objective case). We then extend classical tournament and elite selection strategies into a multi-objective framework which allows classical GP schemes to be readily Pareto-aware. Finally, we introduce the latest extension of the Pareto paradigm which is the melding with ordinal optimization. It appears that ordinal optimization will provide a theoretical foundation to guide algorithm design. Application of these insights has already produced at least a four-fold improvement in the ParetoGP performance for a suite of test problems.

Research in information security is no longer limited to technical issues: human-related issues such as trust and the sense of security are also required by the user. In this paper, we use a Japanese word for such feelings, Anshin; “An” means to ease, and “Shin” is to mind. One feels Anshin when he is free from worry and fear. We try to identify the factors of Anshin so that we can construct a framework of the evaluation of Anshin. We present an initial Anshin model, and report our recent research results from user survey with factor analysis. We derive the following factors from the analysis; 1) user expectation of trust and confidence, 2) satisfaction with user interface and 3) understanding of risk and threats from user experience as well prior knowledge.

The Web represents an information space where the amount of information grows exponentially. This calls for personalized interaction between users and web-based information systems providing information. Current systems provide a certain level of personalization, which allows the user to set up her preferences manually. Improved efficiency of information acquisition can be achieved by personalization based on a user’s particularities used for the adaptation of content or navigation in the information space. A user model that reflects a real user, who requires information provided by an information system, is required for successful personalization. We present an ontology-based approach to user modeling and describe the user model that we designed for a web-based information system aimed at job acquisition. We point out several advantages of the ontology-based approach, namely the sharing of the ontology with other applications and reusability.

Adaptive systems behavior based on user models appear promising, mostly for complex environments such as mixed reality environments (MRE). An MRE comprises a virtual representation of the reality as well as physical objects augmented with virtual features. These objects are coupled with the virtual representation so that they can reflect its changes in real time. The proper design of an MRE and the user models that it implies are crucial for its success, but unfortunately, there are no guidelines for the design of these environments. In this paper we present a methodology for designing user models for MRE as well as for the augmentation of physical everyday objects. The user model describes users’ knowledge in two levels of abstraction: objects manipulation (syntax) and its meaning assigned by a community of practice (semantics).

User interfaces are commonly designed to match the requirements of the so-called ‘typical user’ who represents an abstract generalization of each user of an application. In systems employed by different kinds of users (such as Web systems), the identification of the requirements of their users is very difficult, if not impossible. Our proposal is to avoid the construction of interactive dialogs during the design stage, building them dynamically once the specific cognitive, perceptual and motor requirements of the current user of the system are known: that is, during the execution stage.

Emerging telecommunications architectures, such as IP telephony, offer more signaling flexibility, elaborated feature creation techniques and user preference expression possibilities. The present article focuses on the last point. At our knowledge, no work that presents a thorough analysis of the user preference concept has been reported in the literature. The present article aims at inciting interest to this research axis. It specifies a user preference conceptual model to determine the different concepts that may be subject to user preferences. This conceptual model allows us to categorize the preference expression schemas that are possible to perform on the different conceptual model elements, and specify the architectural and language requirements necessary to implement those preference expression schemas. Examples of new services obtained by the user preference growing expression possibilities are then given. The benefits of the proposed user preference conceptual model are to direct the research work in this area and to provide a theoretical tool to analyze the expressive power of user preference modeling languages.

A new approach to dynamic creation of user interfaces in virtual reality environments is presented. In the proposed approach, templates based on a new virtual reality language X-VRML are used. An X-VRML template consists of the user interface specification and the 3D environment model. Both parts are specified in X-VRML and interpreted by the X-VRML interpreter. A visualized instance of the user interface is used to supply the 3D environment model with initial parameter values that permits to instantiate a 3D scene. In the proposed approach, the specification of the user interface is abstract and does not rely on any particular implementation of visual elements, allowing creation of different user interface implementations according to an actual system architecture, availability of visual component libraries, as well as user needs and preferences.

In market oriented product innovation the users have a central place. Traditionally, the focus has been on fulfilment of their rational and emotional needs, but increasingly societal needs, first of all related to environment, health and safety, have to be taken into consideration.

This paper describes the use of SDL for defining User Models for a test and measurement system whose purpose is to perform load tests of the control unit of an ISDN-PBX with realistic signalling traffic. The generation of realistic signalling traffic is based on the emulation of real users by User Models. The functional behaviour of a User Model is defined by an SDL process using a subset of the SDL syntax. The use of this type of User Models allows, besides the creation of messages according to distribution functions, to process messages received from the system under test and therefore to react appropriately on them, like a real user would do it. Normally several User Models are used together for a load test, so that the generated traffic is a realistic mixture of a variety of different user categories. The User Models are transformed to data structures for execution by the load test system. These data structures are optimized for efficiency in order to reach the target call rate of at least 30000 calls per hour on a standard workstation. The temporal behaviour of the User Models is defined separately by means of Event Generators which send signals to the User Models according to statistical distribution functions.

More effective, efficient and natural human computer or computer mediated human-human interaction will require both automated understanding and generation of multimedia. Fluent conversational interaction demands explicit models of the user, discourse, task and context. It will also require a richer understanding of media (i.e., text, audio, video), both in its use in the interface to support interaction with the user as well as its use in access to content by the user during a session. Multimedia dialogue prototypes have been developed in several application domains including CUBRICON (for a mission planning domain) (Neal and Shapiro, 1991), XTRA (tax-form preparation) (Wahlster, 1991), AIMI (air mission planning) (Burger and Marshall, 1993), and AlFresco (art history information exploration) (Stock et al., 1993). Typically, these systems parse mixed (typically asynchronous) multimedia input and generate coordinated multimedia output. They also attempt to maintain coherency, cohesion, and consistency across both multimedia input and output. For example, these systems often support integrated language and deixis for both input and output. They extend research in discourse and user modeling (Kobsa and Wahlster, 1989) by incorporating representations of media to enable media (cross) reference and reuse over the course of a session with a user. These enhanced representations support the exploitation of user perceptual abilities and media preferences as well as the resolution of multimedia references (e.g. “Send this plane there” articulated with synchronous gestures on a map).

There have been several recent examples of user interface techniques in which the user uses a computational device by physically manipulating the device. This paper proposes that these form an interesting new paradigm for user interface design, Embodied User Interfaces. This paper presents and defines this paradigm, and places it in the evolution of user interface paradigms leading towards the ideal of an invisible user interface. This paper outlines the space of design possibilities in this paradigm, presents a design framework for embodied user interface design, and articulates a set of design principles to guide design.

In most recent Visual Query Systems (VQSs), there is the attempt to address the different needs of various classes of users by developing adaptive interfaces having a fundamental component in the user model module. Within this context, we argue that the temporal issues are a key factor in modeling th user’s interaction, and also in carrying out related usability analysis on VQSs. Hence, we propose a uniform framework for modeling and manipulating several kinds of time-based information related with VQSs, which exploits a graph-based formalism, equipped with visual temporal operators. In this paper, we concentrate on the time-based information which are derived from usability experiments. In particular, we show that the history of the user’s interaction can be accessed and dynamically maintained by using such visual temporal operators. This approach is an effort towards clearly establishing a link between user interaction and modeling and data modeling and querying, allowing a structured and well controlled analysis of VQS usability results.

The user interface is a crucial part of most computer applications. Specially in complex applications, the user interface very often determines the acceptance or the rejection of a program by the user. The design of graphical user interfaces for multi-agent systems is specially difficult due to the particular characteristics of these systems. This difficulty is even greater when we consider the existence of more than one user. In this paper we present a proposal for a general purpose multi-agent multi-user interface. This interface supports a user taxonomy and the ability to continuously learn about the aptitude of each user in order to choose the most correct interlocutor to interact with at each moment. Cooperative decisions and conflict resolution between users are also made possible trough the automatic negotiation of the different solutions envisaged for each problem. For this purpose, a negotiation algorithm for multi-user conflict resolution inspired by the Contract Net Protocol is presented and a specific systems user scoring procedure is proposed.

This paper addresses the specification and verification of hardware interface behaviors using an algebra of communicating timing charts (ACTC for short) whose underlying user model is the well known timing diagram. Terms modeling hierarchical timing charts are built in two steps. First, basic terms corresponding to the leaf charts are built, and then hierarchical terms are constructed using hierarchical operators. The basic terms model elementary behaviors as observed at the interface, and the hierarchical terms allow modeling behaviors at a higher level of abstraction: e.g., modeling an exception handling mechanism. We define an equivalence relation over the chart terms which is preserved by all the language constructs, and give a strategy to decide whether two deadlock-free basic terms are equivalent.