Judy Kay

A fair share scheduler

Central-processing-unit schedulers have traditionally allocated resources fairly among processes. By contrast, a fair Share scheduler allocates resources so that users get their fair machine share over a long period.

Learner Control

This paper describes major trends in learner-adapted teaching systems towards greater learner control over the learning process. In the early teaching systems, the goal was to build a clever teacher able to communicate knowledge to the individual student. Recent and emerging work focuses on the learner exploring, designing, constructing, making sense and using adaptive systems as tools. Correspondingly, systems are being built to give the learner greater responsibility and control over all aspects of the learning, and especially over the learner model which is at the core of user-adaptation. A parallel trend is the growing acknowledgement of the importance of the learners social context. Systems are increasingly being designed for learners working in groups of real or simulated peers. This paper discusses several elements of the shift to greater learner control, with a focus on the implications for learner modelling. The computer may offer the learner a choice of learning tools and companion learners, on-demand learning of various types, control over the elements of the systems and the possibility of controlling the amount of control. Learner control offers promising possibilities for improved learning. At the same time, there are pragmatic issues for achieving the benefits. The paper discusses three of these: the need to evaluate the effectiveness of the emergent learner-controlled systems; problems with learner control; and the need for interoperable and reusable components.

Clustering and Sequential Pattern Mining of Online Collaborative Learning Data

Group work is widespread in education. The growing use of online tools supporting group work generates huge amounts of data. We aim to exploit this data to support mirroring: presenting useful high-level views of information about the group, together with desired patterns characterizing the behavior of strong groups. The goal is to enable the groups and their facilitators to see relevant aspects of the groups operation and provide feedback if these are more likely to be associated with positive or negative outcomes and indicate where the problems are. We explore how useful mirror information can be extracted via a theory-driven approach and a range of clustering and sequential pattern mining. The context is a senior software development project where students use the collaboration tool TRAC. We extract patterns distinguishing the better from the weaker groups and get insights in the success factors. The results point to the importance of leadership and group interaction, and give promising indications if they are occurring. Patterns indicating good individual practices were also identified. We found that some key measures can be mined from early data. The results are promising for advising groups at the start and early identification of effective and poor practices, in time for remediation.Group work is widespread in education. The growing use of online tools supporting group work generates huge amounts of data. We aim to exploit this data to support mirroring: presenting useful high...

Tabletop sharing of digital photographs for the elderly

We have recently begun to see hardware support for the tabletop user interface, offering a number of new ways for humans to interact with computers. Tabletops offer great potential for face-to-face social interaction; advances in touch technology and computer graphics provide natural ways to directly manipulate virtual objects, which we can display on the tabletop surface. Such an interface has the potential to benefit a wide range of the population and it is important that we design for usability and learnability with diverse groups of people.This paper describes the design of SharePic -- a multiuser, multi-touch, gestural, collaborative digital photograph sharing application for a tabletop -- and our evaluation with both young adult and elderly user groups. We describe the guidelines we have developed for the design of tabletop interfaces for a range of adult users, including elders, and the user interface we have built based on them. Novel aspects of the interface include a design strongly influenced by the metaphor of physical photographs placed on the table with interaction techniques designed to be easy to learn and easy to remember. In our evaluation, we gave users the final task of creating a digital postcard from a collage of photographs and performed a realistic think-aloud with pairs of novice participants learning together, from a tutorial script.

Personis: A Server for User Models

A core element of an adaptive hypertext systems is the user model. This paper describes Personis, a user model server. We describe the architecture, design and implementation. We also describe the way that it is intended to operate in conjunction with the rest of an adaptive hypertext system.A distinctive aspect of the Personis user model server follows from our concern for making adaptive systems scrutable: these enable users to see the details of the information held about them, the processes used to gather it and the way that it is used to personalise an adaptive hypertext. We describe how the architecture supports this.The paper describes our evaluations of the current server. These indicate that the approach and implementation provide a workable server for small to medium sized user collections of information needed to adapt the hypertext.

The um toolkit for cooperative user modelling

This paper gives an overview of the um toolkit: the philosophy underlying its design, examples of its use and discussion of the way it deals with some major issues in creating user modelling shells. The um toolkit has been developed to provide support for a variety of cooperative agents. An important element of its cooperativeness is due to its capacity to give users an understanding of their own user models. This paper describes two substantial but very different uses of the toolkit. The first involves a collection of coaching systems that help users learn more about their text editor. Experimental results suggest that the user model is associated with users learning more. The second is a movie advisor that uses a range of tools to construct and refine the user model and to filter a database of movies. Both these systems are built from combining tools in um. The paper describes several of the tools for constructing and refining user models. In addition it describes the user-model viewing tools and the way that these help users ensure their user models are correct. The paper also discusses the two central themes of the um work, the application of a tools approach to the design of a user modelling toolkit and the implications of making the user model accessible to its owner, the person modelled.

Stereotypes, Student Models and Scrutability

Stereotypes are widely used in both Intelligent Teaching Systems and in a range of other teaching and advisory software. Yet the notion of stereotype is very loose. This paper gives a working definition of stereotypes for student modelling. The paper shows the role of stereotypes in classic approaches to student modelling via overlay, differential and buggy models.A scrutable student model enables learners to scrutinise their models to determine what the system believes about them and how it determined those beliefs. The paper explores the ways that scrutable stereotypes can provide a foundation for learners to tune their student models and explore the impact of the student model. Linking this to existing work, the paper notes how scrutable stereotypes might support reflection and metacognition as well as efficient, learner-controlled student modelling.

MOOCs: So Many Learners, So Much Potential ...

Massive open online courses (MOOCs) have exploded onto the scene, promising to satisfy a worldwide thirst for a high-quality, personalized, and free education. This article explores where MOOCs fit within the e-learning and Artificial Intelligence in Education (AIED) landscape.

RECON: a reciprocal recommender for online dating

The reciprocal recommender is a class of recommender system that is important for several tasks where people are both the subjects and objects of the recommendation. Some examples are: job recommendation, mentor-mentee matching, and online dating. Despite the importance of this type of recommender, our work is the first to distinguish it and define its properties. We have implemented RECON, a reciprocal recommender for online dating, and have evaluated it on a large dataset from a major Australian dating website. We investigated the predictive power gained by taking account of reciprocity, finding that it is substantial, for example it improved the success rate of the top ten recommendations from 23% to 42% and also improved the recall at the same time. We also found reciprocity to help with the cold start problem obtaining a success rate of 26% for the top ten recommendations for new users. We discuss the implications of these results for broader uses of our approach for other reciprocal recommenders.

PersonisAD: distributed, active, scrutable model framework for context-aware services

PersonisAD, is a framework for building context-aware, ubiquitous applications: its defining foundation is a consistent mechanism for scrutable modelling of people, sensors, devices and places. This paper describes the PersonisAD features for supporting distributed models with active elements which can trigger when relevant events occur. This framework makes it possible to quickly create new context-aware applications. We demonstrate the power of the framework by describing how it has been used to create two context aware applications: MusicMix which plays music based on the preferences of the people in the room; MyPlace, which informs people of relevant details of the current environment. Major contributions of this work are: the PersonisAD framework which provides a powerful and consistent means to respond to significant changes in the models of people, sensors, devices and places; support for distributed models and associated resource discovery; two applications that illustrate the power of PersonisAD.