CHS: Small: Collaborative Research: Mobile Language-Based Aids for Intelligent Decisions

Auburn University, Auburn AL

Investigators

Abstract

This project will create a solid science base for computer-based conversational decision systems, by integrating psychological decision theories with an experiential understanding of human decision-making, having the potential to impact many domains. People make mundane and critical consumption decisions daily, ranging from what air conditioning filters to buy to which health insurance plan to enroll. Rational choice theory predicts that consumers can make choices to maximize their desired attributes. However, human cognitive limitations coupled with decision complexities often require people to develop preferences quickly and employ multiple decision strategies to arrive at a choice. Sub-optimal consumption decisions hurt the individual, the family, the economy, and the nation. This project will enhance people's abilities to make informed decisions on-the-go, when they lack domain knowledge, time and cognitive resources for optimal decision-making. These benefits are particularly relevant to aging users, those with cognitive, perceptual and motor impairments, and those functioning in multitask environments. Past research on decision aids has focused on single-decision strategies using invariant algorithms, falling short of solving real decision problems, which are inherently contingent on the decision-maker and task. Further, benefits of a mobile, conversational setting have been under-employed in developing decision aids. The goal of this project is to design, develop, and evaluate conversational, mobile decision-aids (MODA) for consumption decisions based on adaptive, intelligent information retrieval and decision strategy use. The three main scientific objectives and methods are: 1. Capture rich spoken dialog between consumers and service associates as they go through different types of decision-making strategies in a physical store, using observational studies. 2. Develop a decision inference and dialog management model that (a) uses inputs from spoken consumer interactions, (b) applies decision inference rules to predict the consumer's decision strategy, (c) tailors MODA's subsequent dialog to their decision strategy, and (d) informs the consumer's decision-making process with data retrieved from product specifications and reviews. 3. Conduct field experiments to assess MODA effectiveness with respect to cognitive, social, and decision outcomes among varied decision-contexts and user characteristics.