An Attention-Based Entity Linking Method for Chinese Knowledge Base Question Answering System

Entity linking for knowledge base question answering is the process of identifying entities in the question and linking them to knowledge base entities. There are some problems in entity linking of Chinese knowledge base question answering, such as missing context information, unclear word boundaries, and too many colloquialized words. We propose an attention-based entity linking method. In the candidate entity generation stage, we use various ways to generate candidate entities. In the candidate entity disambiguation stage, attention mechanism is introduced to highlight the features that are more critical for entity disambiguation. We conduct experiments on public data sets, and the results show the effectiveness of our method.