Automated Exploration of Conversational Agents for the Synthesis of Testing Profiles
This program is tentative and subject to change.
Conversational agents – or chatbots – are increasingly being used to access all sorts of services, like citizen services in city halls, customer support, or shopping. Moreover, recent advances in generative artificial intelligence are prompting the integration of conversational assistants into many applications, like programming IDEs, office automation software, or operating systems. Given the prominence of these agents, their correctness is a rising concern. However, automated and robust testing techniques for conversational systems are still needed.
In this paper, we present a technique for extracting a model of a deployed chatbot (i.e., treated as a black-box) through the automated exploration of its functionality via Large Language Models. This model is used for automated testing by generating testing conversation profiles, which a user simulator employs to conduct focused conversations with the chatbot-under-test. We describe our tool support, and report on an evaluation showing that our exploration technique can accurately model the chatbot-under-test, and the subsequent testing can discover existing errors in the chatbot.
This program is tentative and subject to change.
Fri 19 SepDisplayed time zone: Athens change
14:00 - 15:30 | LLMs and Agent-Based TestingGeneral Track at Atrium C Chair(s): Jørn Eirik Betten Simula Research Laboratory; Oslo Metropolitan University | ||
14:00 30mTalk | Reverse Engineering for Input Modeling: Input Parameter Model Inference from Network Traces General Track Manuel Leithner SBA Research, Salzburg University of Applied Sciences, Dimitris E. Simos Salzburg University of Applied Sciences, Paris LodronUniversity of Salzburg | ||
14:30 30mTalk | Automated Exploration of Conversational Agents for the Synthesis of Testing Profiles General Track Iván Sotillo del Horno Universidad Autónoma de Madrid, Alejandro del Pozzo Universidad Autónoma de Madrid, Esther Guerra Universidad Autónoma de Madrid, Juan de Lara Autonomous University of Madrid Pre-print Media Attached | ||
15:00 30mTalk | Extracting Threats from System Descriptions with LLMs - Comparing One and Two Agents Strategies General Track |