An Entity Linking Algorithm Derived from Graph Convolutional Network and Contextualized Semantic Relevance

In the era of big data, a large amount unstructured text data springs up every day. Entity linking involves relating mentions found in texts to corresponding entities, which stand for objective things real world, knowledge base. This task can help computers understand semantics correctly. Although there have been numerous approaches employed research such as this, some challenges are still unresolved. Most current utilize neural models learn important features entity and mention context. However, topic coherence among referred entities is frequently ignored, leads clear preference popular but poor accuracy less ones. Moreover, graph-based face much noise information high computational complexity. To solve problems above, paper puts forward an algorithm derived from asymmetric graph convolutional network contextualized semantic relevance, make full use neighboring node well deal with unnecessary graph. The vector candidate obtained by continuously iterating aggregating nodes. relevance model symmetrical structure that designed realize deep measurement between entities. experimental results show proposed fully explore topology dramatically improve effect compared baselines.