Invited speakers
The following keynote speakers will give invited presentations at MLMI 2008:
HCI2: Human-Centered Intelligent Human-Computer Interaction
Maja Pantic, Department of Computing, Imperial College London and University of Twente
A widely accepted prediction is that computing will move to the background, weaving itself into the fabric of our everyday living spaces and projecting the human user into the foreground. To realize this prediction, next-generation computing should develop anticipatory user interfaces that are human-centred, built for humans, and based on naturally occurring multimodal human communication. They should transcend the traditional keyboard and mouse to include natural interactive functions including understanding and emulating both human communicative signals such as facial expressions, body gestures, and vocal outbursts and human behaviors such as affective and social signaling. This talk discusses how far are we to the goal of human-centred computing and HCI2 computer interfaces that can understand multimodal human communicative behaviour.
Communicating with a distinctive embodied conversation agent
Catherine Pelachaud, IUT de Montreuil and INRIA Rocquencourt
Embodied conversational agents, ECAs, are software entities capable of autonomously communicating with users through verbal and nonverbal means. As such, ECAs are endowed with the ability to display human-like nonverbal behaviours in order to convey information reflecting their mental and emotional states. People differ in their communicative behaviours; some may have a large gestural space, other may gaze more, and so on. We have developed an algorithm that encompasses these differences in behaviour communication. In our presentation we will describe the research we have undertaken to create a distinctive conversational agent, named Greta.
Measuring interactivity on search engines is hard, unless you're Yahoo, Microsoft, or Google
Mark Sanderson, Department of Information Studies, University of Sheffield
The field of information retrieval is rightly proud of its long history of standardizing evaluation methodologies and sharing data sets to help researchers evaluate their particular search algorithms. This sharing of test sets started in the 1960s. The methodologies devised back then have stood the test of time and are still in use today. The evaluation that can be done with these data sets, known as test collections, is however, limited. Specifically interactivity and browsing in search cannot be studied with a test collection. In this talk I will describe some of the attempts by retrieval researchers to adapt the test collection methodology to make it more "realistic". I will cover a brief history of interactive evaluations; testing in the context of adaptive search and the use of query logs in that work; as well as more recent work on measuring the link between test collection results and user satisfaction when using a search engine.
The "Giraffe" - A Mobile Telepresence Physical Avatar
Stephen Von Rump, HeadThere Inc., San Francisco, CA
HeadThere Corporation has developed a mobile telepresence device called the "Giraffe," that allows a person to project themselves into a venue and move about and communicate without requiring those local to the device to do anything. The device is a remotely controlled, human-height physical avatar integrated with a videoconferencing system that can be accessed via a simple laptop interface over the Internet. A remote person with no prior training can navigate the Giraffe down hallways, through doorways and around tables and chairs, look around via a pan/tilt/zoom camera, and be seen and heard in real time via a life-size portrait view from their own webcam. The remote person can even adjust the device's height to "sit" at a table or bedside. This presentation explores several industry opportunities for the Giraffe, addressing the technical and business challenges of each, along with a proposed product roadmap. The presentation is supported by a live demonstration of the device.