A Comparative Evaluation of Natural Language and Dashboard Interfaces for Visualizing Real-Time Monitoring Data

Abstract

Natural Language Interfaces (NLIs) are emerging as an alternative to Dashboard Interfaces for data visualization, allowing users to formulate queries using conversational input rather than structured commands. In a controlled study (N=24) we compare an NLI-driven chatbot and a commercial visualization dashboard (RealFishPro) for a set of randomized analytics tasks involving real-time monitoring data. We find no difference in System Usability Scale (SUS) scores and no difference in task accuracy scores between interface conditions. NASA TLX scores show significantly higher mental and temporal demand when using the chatbot vs. the dashboard, and the chatbot yielded significantly higher task times overall. This pattern shows that while NLIs are flexible, they often impose greater cognitive effort and slower interaction compared to dashboards. Participant feedback indicated complementary strengths: NLIs were praised for simplicity and adaptability, dashboards for precision and clarity. These findings suggest that hybrid solutions integrating natural language and traditional interfaces could enhance data exploration and decision-making.