Conversational case-based reasoning in self-healing and recovery

Self-healing and recovery informed by environment knowledge (SHRIEK) is an autonomic computing approach to improving the robustness of computing systems. Case-based reasoning (CBR) is used to guide fault diagnosis and enable learning from experience, and rule-based reasoning to enable fault remediation and recovery informed by environment knowledge. Focusing on the role of conversational CBR (CCBR) in the management of faults that rely on user interaction for their detection and diagnosis, we present a hypothesis-driven approach to question,,election in CCBR that aims to increase the transparency of CCBR dialogues by enabling the system to explain the relevance of any question the user is asked. We also present empirical results which suggest that there is no loss of problem-solving efficiency in the approach. Finally, we investigate the effects of the environment awareness provided by autonomous information gathering in SHRIEK on the efficiency of CCBR dialogues.