Natural Language Dialogue group

The VIGOR project is focused on creating different types of virtual characters to inhabit and virtual worlds and provide a richer experience for human visitors to such spaces.

Online worlds hold great promise for purposes as diverse as education, healing and socializing. Virtual worlds are persistent, easily accessible, and always available through one’s connected computer. These worlds provide opportunities to build dedicated spaces that serve specific populations.
Even though people are not physically co-located they can view the same virtual objects and information, get to know each other, and work and learn together within the shared virtual space. There are a few challenges to realizing the promise of virtual worlds. If someone enters the virtual world for the first time but is not able to find anyone or anything interesting to interact with, they may proclaim the visit boring or not meaningful. Thus they are not exposed to the value of the virtual world, even though it may be quite relevant to their interests and needs. Also, virtual worlds usually require some instruction to bring participants up to speed with navigation and communication methods. This is currently handled by “orientation areas” where a new person has to perform, on their own, a series of tasks designed to get them started. If they get confused, however, this situation often results in failure. It is much easier (and much more successful) to learn to use technology with someone there to answer questions and guide you as to what you can do and what opportunities are available.
Having an expert present who can guide the interaction toward productive experiences is often key to participants realizing the potential of virtual environments. These issues argue for having a permanent staff within the virtual environment, so users can be supported whenever they visit. However it can be prohibitively costly to have a full-time human staff, and many of the interactions will be fairly routine – users requiring the same kind of question-answering and need the same guidance for what is available within the environment, how to navigate, engage in activities and communicate.

In this project, we take the first steps towards the creation of several types of intelligent virtual humans in Second Life and Active Worlds:
(1) Virtual question answering characters (e.g. Ahmad Wani) that can answer questions about relevant domains or redirect questions that are not on topic.

(2) Virtual guides, who can pro-actively initiate conversation with newcomers, find out their interests and guide or teleport them to the most relevant parts of the virtual environment, e.g LT Moleno in Second Life.

(3) A prototype of a virtual storyteller or experience leader, who can narrate a particular event, some of which can be seen in the virtual world, and also take questions or tailor the presentation to the specific interests of the audience. (e.g. Jumpmaster Neiro in Second Life, or the Staff Ride Guide in Activeworlds). These characters can be augmented with tutoring capabilities, such as after-action reviews.

(4) Culture-specific characters who are configured to represent and act according to models of behavioral norms from different cultures.

(5) Virtual team-members, who can help train group tasks with only a single individual or sub-group.