2013 · 161 citations · 13 references
ComputingComputer ScienceArtificial IntelligenceMachine LearningNatural Language ProcessingConversational InterfacesConversational User InterfacesDialogue SystemsProbabilistic Machine LearningAccuracyDialog SystemsEnd-to-end Dialogue SystemsEvaluation Of Dialogue SystemsSpoken Dialogue SystemsUser GoalsUser Goals Modeling
The authors propose a generic dialogue state tracker that uses simple, domain‑independent rules to maintain beliefs over user goals, aiming to maximize information extracted from the dialogue itself. The tracker applies basic probability operations to observed system actions and partially observable user acts, without relying on external resources or training data. In the Dialog State Tracking Challenge, the method attains hypothesis‑accuracy comparable to machine‑learning baselines and highlights the relative strengths and weaknesses of statistical approaches.
This paper presents a generic dialogue state tracker that maintains beliefs over user goals based on a few simple domainindependent rules, using basic probability operations. The rules apply to observed system actions and partially observable user acts, without using any knowledge obtained from external resources (i.e. without requiring training data). The core insight is to maximise the amount of information directly gainable from an errorprone dialogue itself, so as to better lowerbound one’s expectations on the performance of more advanced statistical techniques for the task. The proposed method is evaluated in the Dialog State Tracking Challenge, where it achieves comparable performance in hypothesis accuracy to machine learning based systems. Consequently, with respect to different scenarios for the belief tracking problem, the potential superiority and weakness of machine learning approaches in general are investigated.
13
The Dialog State Tracking Challenge
J. D. Williams, Antoine Raux, Deepak Ramachandran et al. · 2013 · 332 citations