Self-Improvement Programming for Temporal Knowledge Graph Question Answering

arXiv:2404.01720v1 Announce Type: new
Abstract: Temporal Knowledge Graph Question Answering (TKGQA) aims to answer questions with temporal intent over Temporal Knowledge Graphs (TKGs). The core challenge of this task lies in understanding the complex semantic information regarding multiple types of time constraints (e.g., before, first) in questions. Existing end-to-end methods implicitly model the time constraints by learning time-aware embeddings of questions and candidate answers, which is far from understanding the question comprehensively. Motivated by semantic-parsing-based approaches that explicitly model constraints …

abstract arxiv challenge constraints core cs.cl graph graphs improvement information knowledge knowledge graph knowledge graphs lies multiple programming question question answering questions self-improvement semantic temporal type types understanding