Presentation Information
[1Yin-B-53]Design guidelines for sales support chatbots based on expert question formulation
〇TAKATO KANAYAMA1, YUTA NAKAGAWA1, RISA IGUCHI1 (1. Matsumoto Trading Co.,Ltd.)
Keywords:
Visualization of specialized skills,Sales Support,Large-scale language models,Chatbots
Even in environments where users have access to the same information sources, the quality of information obtained varies depending on the user. Similar challenges exist in tasks requiring specialized knowledge, such as responding to technical inquiries from customers in sales at a cosmetics raw materials trading company. In this study, in order to identify effective strategies for developing sales support chatbots, we focused on the differences in vocabulary and depth of questions based on experience, which have been pointed out in previous research, and examined the impact of these factors. Results of Q&A analysis and information search experiments within the same organization showed no significant differences in vocabulary used based on experience, but experienced users were superior in their ability to reframe problems in a way that can be solved. These results suggest that dialogue design that emphasizes the depth of questions is important for utilizing LLM in tasks requiring specialized knowledge.