Custom Accessibility-Based CCBR Question Selection by Ongoing User Classification

Question selection in Conversational Case-Based Reasoning (CCBR) is traditionally guided by the discriminativeness of questions, to minimize dialog length for retrieval. However, users may not always be able or willing to answer the most discriminative questions. This paper presents Accessibility Influenced Attribute Selection Plus (AIAS+), a method for customizing CCBR question selection to reflect the types of questions the user is likely to answer. Given background knowledge about response probabilities for different questions by different user groups, AIAS+ performs ongoing classification of new users, based on the questions they choose to answer, uses the classifications to predict the likelihood of the user answering particular questions, and applies those predictions to guide question selection. In addition, its question selection process balances questions’ information gain against their potential to aid user classification, to enable better selection of following questions. Experiments with simulated users show improvement over three alternative methods. Experiments in synthetic domains illuminate the domain characteristics under which the method is expected to be effective.