AIAI Seminar - 24 January 2023 - Talk by Ricky Zhu

Title: Continuous Evolution of Probabilistic Knowledge Base (2^nd Year Review)

Abstract: Knowledge Base (KB) or Knowledge Graph (KG) can empower many intelligence applications, such as question answering and fault analysis. However, Knowledge bases are not available in many specific domains, and constructing a knowledge base manually is very expensive. My PhD program aims at automatically constructing a probabilistic knowledge base for a running software system (e.g., a telecom network) by extracting knowledge from system documentation and system logs . Then we utilise the continuously extracted knowledge from the logs to update and revise the existing knowledge, making the knowledge base self-evolving, i.e. correcting potential errors and capturing the changes of a running software system. In this 2^nd Year Review, I will briefly recap what I have done in the first year and report what I did in the second year. An updated work plan and risk assessment will be presented as well.