A natural language technology-enhanced mobile sales assistant for in-store shopping situations

Sales talks between customers and sales personnel are efficient and preferred means for exchanging information that is relevant for purchase decisions on non-commodity products. Dialogs used in sales talks are governed by complex and in many respects conflicting intentions on both sides. While previous Decision Support Systems (DSS) are designed by the principle of congruent intentions of communication partners, we present an approach that extends this by congruent and opposing intentions of communication partners. We use a design methodology for dialog-based purchase DSS that use Natural Language Technologies (NLT) for dynamically creating question-answer-based sales dialogs. It is first shown how dialog schemata are obtained by a field study and evaluated by subjects. In the second part, these schemata are integrated in a Natural Language Technology-enhanced Mobile Sales Assistant (NLT-MSA). The role of NLT-MSA is to take the position of a sales person with the task to balance congruent and conflicting intentions during sales dialogs. Results of a lab experiment (n=54) are discussed by which the use of a NLT-MSA prototype in sales situations were tested. As part of this study, test persons rated application domains for NLT-MSA that will guide future field experiments in the large.