Zero-shot Entity Linking with Efficient Long Range Sequence Modeling

This paper considers the problem of zero-shot entity linking, in which a link in the test time may not present in training. Following the prevailing BERTbased research efforts, we find a simple yet effective way is to expand the long-range sequence modeling. Unlike many previous methods, our method does not require expensive pre-training of BERT with long position embeddings. Instead, we propose an efficient position embeddings initialization method called Embedding-repeat, which initializes larger position embeddings based on BERT-Base. On Wikia's zero-shot EL dataset, our method improves the SOTA from 76.06% to 79.08%, and for its long Figure 1: Only models with large ERLength can solve data, the corresponding improvement is from this entity linking problem because only they can get 74.57% to 82.14%. Our experiments suggest valuable critical information in the mention contexts the effectiveness of long-range sequence modeling and entity description.