CFXGBoost: Topic phrase extraction based on context features and XGBoost for knowledge base question answering
Han Han, Fengyu Yang, Jiuming Huang, Bin Zhou · 2017
Topic phrase extraction is a core component for many NLP(Natural Language Processing) applications including Knowledge Base Question Answering(KBQA). In this paper, we propose a topic phrase extraction method based on context features and XGBoost(CFXGBoost) to detect the topic phrase given an input question. Our method first finds all of the phrases which appear in the question through the pre-constructed phrase dictionary, then extracts rich context features of the phrases to train a XGBoost model to detect the topic phrase in the question. Experimental results evidence that our approach achieves outstanding topic phrase extraction performance in KBQA system.