Human-Centered Distributed Conversational Modeling: Efficient Modeling of Robust Virtual Human Conversations

Abstract

Abstract. Currently, applications that focus on providing conversations with virtual humans require extensive work to create robust conversational models. We present a new approach called Human-centered Distributed Conversational Modeling. Using this approach, users create conversational models in a distributed manner. To do this, end-users interact with virtual humans to provide new stimuli (questions and statements), and domain-specific experts (e.g. medical/psychology educators) provide new virtual human responses. Using this process, users become the primary developers of conversational models. We tested our approach by creating an example application, Virtual People Factory. Using Virtual People Factory, a pharmacy instructor and 186 pharmacy students were able to create a robust conversational model in 15 hours. This is approximately 10% of the time typical in current approaches and results in more comprehensive coverage of the conversational space. In addition, surveys demonstrate the acceptability of this approach by both educators and students.

High Score! - Motivation Strategies for User Participation in Virtual Human Development

“…Thus, through iterations of user interactions and subsequent expert reviews, the corpus of the VH is progressively developed. The HDCM method was observed to take only 10% of the time taken for conversational modeling by previous approaches [4].…”

“…VPF is a web-based application that allows users to create and interact with VHs [4]; the VHs created using VPF conduct natural language conversations using a corpus retrieval approach.…”

“…The time taken for corpus development can be reduced by using a technique called Human-Centered Distributed Conversational Modeling (HDCM) [4]. In HDCM, novice users conduct interactions with the VH during the VH corpus development phase and pose questions (stimuli) to the VH.…”

“…After 27 iterations of different users interacting followed by expert reviews of the accumulated stimuli, the corpus now has 118 responses and 446 stimuli. The HDCM model saves time [4] but shifts much of the corpus development work from the expert to the novice. This leaves domain experts with the problem of motivating enough novice users to interact with the VH during the corpus development phase.…”

“…In the past, providing monetary compensation and extra credit have been effective motivational techniques for increasing user participation [3] [4], however, these motivational techniques are not always applicable. For example, providing monetary compensation can be costly and may bias participants.…”