Testing whether a probability distribution is compatible with a given Bayesian network is a fundamental task in the field of causal inference, where Bayesian networks model causal relations. Here we consider the class of causal structures where all correlations between observed quantities are solely due to the influence from latent variables. We show that each model of this type imposes a certain signature on the observable covariance matrix in terms of a particular decomposition into positive semidefinite components. This signature, and thus the underlying hypothetical latent structure, can be tested in a computationally efficient manner via semidefinite programming. This stands in stark contrast with the algebraic geometric tools required if the full observable probability distribution is taken into account. The semidefinite test is compared with tests based on entropic inequalities.

Personal communication between User-Centered Design (UCD) specialists and developers is important for communicating user needs and design solutions in agile development. In distributed projects where opportunities for personal communication are limited, the design documentation is an important surrogate. This study has investigated the perceived effectiveness of boundary objects in a distributed agile team, and their role in communicating target user needs. Six in-depth interviews with UCD specialists showed that the boundary objects rarely communicate underlying needs of the users but rather focus on interaction with the system that is being developed. The used boundary objects also do not work as stand-alone deliverables; they need to be explained and elaborated. Making the boundary objects comprehensive enough to be stand-alone is seen as too time consuming and not worth the effort. For agile projects with distributed teams, this creates hand-over and follow-up problems.

To evaluate and develop a service supported by an IT (information technology) system the intention to use the future service should be in focus. The technology acceptance model (TAM) and the theory of planned behaviour (TPB) can both provide knowledge about users' intention to use a service, making them good models to base formative decisions on. Unlike TAM, TPB is concerned with specific information related to the service context, and provide knowledge about what makes IT useable. We used an adapted version of the TPB as the foundation for a formative service evaluation technique called F-SET. We applied the F-SET to a case where two subsequent versions of a service prototype were evaluated. The first prototype was a description of the service supported by Hi-Fi design sketches showing what a web-based meal planning tool could look like. The second prototype consisted of both service processes and the web-based meal planning tool. To find relevant factors that influence future use of such a service, a survey of 28 informants was conducted with the first prototype. The second prototype involved five families who used the service for two weeks. The feedback provided by the families, based on the factors identified in the pre-test, influenced the future direction of the service development. Feedback from the informants was distributed between the service and the IT system, and the most common factors that influence the intention to use the service were time, price, usefulness and availability. Feedback included both positive and negative comments, as well as bugs and suggestions for improvements. We discuss potential improvements and what kind of information to expect from the different constructs of the TPB.

The introductory course on human-computer interaction (HCI) and usable systems, taught early for the computer science and computer engineering education programs, has suffered from bad course evaluations. To investigate the students' beliefs and values in relation to the subject matter and the course, focus groups were used. The results indicate challenges with a lack of maturity and motivation among the students and that they wrongly perceive the subject matter as trivial. An approach for dealing with the challenges in the form of a course re-design is presented and implemented. The evaluation of the new course showed that it worked well and was greatly appreciated.

In times of ageing populations and shrinking care resources, electronic assistive technology (EAT) has the potential of contributing to guaranteeing frail older people a continued high quality of life. This paper provides users and designers of EAT with an instrument for choosing and producing relevant and useful EAT applications in the form of a functional design space. We present the field study that led to the design space, and give advice on using the tool.

User support is an important element in reaching the goal of universal usability for Web information systems. Recent developments indicate that human involvement in user support is a step towards this goal. However, most such efforts are currently being pursued on a purely intuitive basis. This, empirical findings about the role of human assistants are important. In this paper we present the findings from a field study of a general user support model for Web information systems. We show that integrating human assistance into Web systems is a way to provide efficient user support. Further, this integration makes a Web site more fun to use and increases the user's trust in the site. The support also improves the site atmosphere. Our findings are summarised as recommendations and design guidelines for decision-makers and developers Web systems.

This paper explores the role of user modelling in live help systems for e-commerce web sites. There are several potential benefits with user modelling in this context: 1) Human assistants can use the personal information in the user models to provide the users with efficient support tailored to their personal needs; 2) Assistants can be more comfortable in their supporting role; 3) Consultation resources can be saved, and thus, financial savings can be made for the e-commerce company. A user modelling approach has been implemented and deployed in a real web environment as part of a live help system. Following the deployment we have analysed consultation dialogue logs and answers to a questionnaire for participating assistants. The initial results show that assistants consider user modelling to be helpful and that consultation dialogues can be an important source for user model data collection.