Improving Question Answering over Knowledge Graphs Using Graph Summarization

Question Answering (QA) systems over Knowledge Graphs (KGs) (KGQA) automatically answer natural language questions using triples contained in a KG. The key idea is to represent and entities of KG as low-dimensional embeddings. Previous KGQAs have attempted Graph Embedding (KGE) Deep Learning (DL) methods. However, KGEs are too shallow capture the expressive features DL methods process each triple independently. Recently, Convolutional Network (GCN) has shown be excellent providing entity GCNs inefficient because treat all relations equally when aggregating neighbourhoods. Also, problem could occur previous KGQAs: most cases, often an uncertain number answers. To address above issues, we propose graph summarization technique Recurrent Neural (RCNN) GCN. combination GCN RCNN ensures that embeddings propagated together with relevant question, thus better proposed can used tackle issue cannot In this paper, demonstrated on common type questions, which single-relation questions. Experiments provide results compared significantly improves recall actual answers