With the rise of powerful mobile devices and the broad availability of computing power, Automatic Speech Recognition is becoming ubiquitous. A flawless ASR system is still far from existence. Because of this, interactive applications that make use of ASR technology not always recognize speech perfectly, when not, the user must be engaged to repair the transcriptions. We explore a rational user interface that uses of machine learning models to make its best effort in presenting the best repair strategy available to reduce the time in spent the interaction between the user and the system as much as possible. A study is conducted to determine how different candidate policies perform and results are analyzed. After the analysis, the methodology is generalized in terms of a decision theoretical framework that can be used to evaluate the performance of other rational user interfaces that try to optimize an expected cost or utility.

In recent years the development of mobile devices has made great progress and today they account for a large proportion of the internet usage. This trend poses new demands on web developers as more companies and product owners want to customize their services and products for this new market. In this study we examine the pros and cons of the strategies desktop-first and mobile-first in the development of adaptive interfaces for the web. The study aims to create guidelines for situations in which each strategy fits well. The purpose of the study is to create an understanding of how these strategies can help developers and product owners to develop adaptive interfaces based on different goals and objectives. The study was conducted through the development of two high-fidelity prototypes for the redesign of an existing website. A literature review was also performed to create a deeper understanding of the subject. Our conclusion showed that mobile-first was preferable when building a website from scratch. However, there are cases when desktop-first may be preferable, for example, in the redesign of an existing website where the foundation is already finished. Furthermore, the study has led us to the realization that there are many purposes of different websites making the subject complex. Therefore, the developer should always make an evaluation of the product owner's goals and objectives in order to choose the right strategy.

User interface design and development for ubiquitous software applications is challenged by the presence of varying contexts. Context comprises of user'­s computing platform, the environment in which the user is interacting with the application and user characteristics which comprise of user's behavior during interaction and user preferences for interface display and interaction. We present a framework for adaptive user interface generation where adaptation occurs when context changes. This framework introduces three new concepts. First, formalization for representing context is introduced. Our design of context specification is unique since it reflects the association of context with level and nature of user interface adaptation. Secondly, user interface generation life cycle is studied and we define a context model on top of task model to introduce the contextual conditions into user interface generation process. Using the context model, user interface designer can specify contextual requirements and its effect on the user interface. Third, context aware adaptation of user interfaces is achieved by mapping context specifications to various levels of user interface generation life cycle. We designed a specification language called rule specification using which the user interface designer can specify the mapping. With the new design of context representation, context model, and rule specification, we demonstrate how changes in contexts adapts task model which in turn adapts the user interface. / Master of Science

Most computer interfaces are designed in a one-size-fits-all fashion, which does not account for individual differences in abilities and preferences. Some computer users thrive with one software application while another user may struggle to use the same software. Some people tend to perform very well amidst distraction whereas others have a difficult time concentrating on a primary task when distracting agents are present. Much work has been done in quantifying a person's performance, but it has typically been difficult to quantify how difficult a task was for a person to perform. This thesis looks into the stress exhibited by various computer users while performing tasks in both their preferred and non-preferred modalities. The paper surveys the current physiological methods for analyzing human stress and delineates the hardware and software design and implementation of some of these methods. The physiological data-collecting hardware and software were deployed to collect physiological samples from test subjects engaging in memorization and recollection tasks in both an undistracted and a distracted setting. An analysis of the data shows the correlation between preferred modality and performance of tasks in that modality and other modalities. This analysis also shows the correlation between user arousal level and performance with and without distraction. Individual differences are considered by normalizing the physiological data collected for each subject prior to comparison with other subjects. The work presented herein gives insight into the individual differences of various types of computer users and is a precursor to work in adaptive user interface technology. / Master of Science

To a large extent, business-customer interactions are acted out on digital meeting places. When the possibilities for businesses to engage in face-to-face interactions decrease, relationship building and customer service becomes more of a challenge. Digital services are easily duplicated by competitors and with standardization of interfaces and products, customers tend to switch more frequently between providers. One area where the creation and maintenance of loyal customers appears to be highly relevant is the domain of e-banking. Studies have shown that by personalizing the experience for the user, customer loyalty can be enhanced. Existing methods of interface adaptation shifts the responsibility for the resulting user experience design from the designer to either the user or the system. However, research shows that handing over responsibility for the design to the user can damage the user experience. Furthermore, we argue that as long as computers cannot translate the meaning of what a human communicates, and understand the motivation that lies behind her actions, human designers and researchers need to own the responsibility for designing user experiences. Responsive web design differ from the concept of user responsiveness in the way that it is not truly responsive to the user but to the technical device that is used. Following a design science research methodology, this paper presents the development of a conceptual framework for user responsive user experience design (URUXD) that aims to strengthen the bond between user and provider by enabling a more relevant and personalized user experience. The conceptual framework introduces a way to design user responsive information systems that could be useful in domains where the user audience is large and diverse, as in the case of e-banking. A personalized user experience is enabled by transcending the current use of personas as design tools to also involve them in categorizing real-time users through the use of personas as mapping tools. Multiple persona sets are incorporated in the framework which gives the user experience designer the possibility of designing a holistic user experience for each persona set. The framework thus enables the incorporation of multiple GUI designs in an information system that is user responsive, without the risk of violating usability principles. / Interaktionen mellan företag och kund sker nuförtiden oftast på digitala mötesplatser. När möjligheten för företag att träffa kunden öga mot öga minskar blir det en utmaning att skapa nära affärsrelationer och förmedla bra kundservice. Med konkurrenter som enkelt kopierar digitala tjänster och med en standardisering av gränssnitt och tjänster tenderar kunder att oftare byta leverantör. Inom e-banking framstår därför skapande och upprätthållande av lojala kunder som högst relevant. Genom att personifiera användarupplevelsen kan kundlojaliteten förbättras. Befintliga metoder för gränssnittsadaption lämnar över ansvaret för den resulterande designen från designern till antingen användaren eller systemet. Men, om ansvaret för designen tilldelas användaren kan resultatet bli i en skadad användarupplevelse. Så länge datorer inte kan översätta meningen bakom vad en människa kommunicerar eller skapa en förståelse för en användares bakomliggande motivation till varför hen utför handlingar, måste mänskliga designers inneha ansvaret för designen av användarupplevelsen. Vidare skiljer sig responsiv webbdesign från konceptet användarresponsivitet i meningen att responsiv webbdesign inte är direkt responsiv mot användaren utan snarare mot den tekniska apparat som används. Genom att följa en design science forskningsmetodik utvecklades ett konceptuellt ramverk för design av användarresponsiva användarupplevelser (user responsive user experience design (URUXD)). Målet var att stärka bandet mellan användare och leverantör genom att möjliggöra en mer relevant och personifierad användarupplevelse. Det konceptuella ramverket introducerar ett sätt att designa användarresponsiva informationssystem vilket kan vara användbart i domäner där användargruppen är stor och heterogen, vilket är fallet för e-banking. En personifierad användarupplevelse möjliggörs genom att utöka det befintliga användningsområdet för designverktyget personas till att även inkludera dem som mappningssverktyg för att kategorisera användare i realtid. Multipla persona sets införlivas i ramverket vilket skapar möjlighet för designern att skapa en holistisk användarupplevelse för varje enskilt persona set. Det konceptuella ramverket möjliggör därigenom för multipla gränssnittdesigns för ett informationssystem som därmed blir användarresponsivt, utan att underminera principer för användbarhet.

User Modeling and Machine Learning for User Modeling have both become important research topics and key techniques in recent adaptive systems. One of the most intriguing problems in the `information age´ is how to filter relevant information from the huge amount of available data. This problem is tackled by using models of the user´s interest in order to increase precision and discriminate interesting information from un-interesting data. However, any user modeling approach suffers from several major drawbacks: User models built by the system need to be inspectable and understandable by the user himself. Secondly, users in general are not willing to give feedback concerning user satisfaction by the delivered results. Without any evidence for the user´s interest, it is hard to induce a hypothetical user model at all. Finally, most current systems do not draw a line of distinction between domain knowledge and user model which makes the adequacy of a user model hard to determine. This thesis presents the novel approach of conceptual user models. Conceptual user models are easy to inspect and understand and allow for the system to explain its actions to the user. It is shown, that ILP can be applied for the task of inducing user models from feedback, and a method for using mutual feedback for sample enlargement is introduced. Results are evaluated independently of domain knowledge within a clear machine learning problem definition. The whole concept presented is realized in a meta web search engine called OySTER.