Zero-shot Entity Linking with Efficient Long Range Sequence Modeling

TL;DR

This paper introduces an efficient method for zero-shot entity linking that enhances long-range sequence modeling without retraining BERT, achieving state-of-the-art results on Wikia datasets.

Contribution

Proposes Embedding-repeat, a novel position embedding initialization method that improves long-range sequence modeling in BERT-based zero-shot entity linking without additional pre-training.

Findings

Achieves SOTA accuracy of 79.08% on Wikia's zero-shot EL dataset.

Improves long data accuracy from 74.57% to 82.14%.

Demonstrates effectiveness without retraining BERT.

Abstract

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 BERT-based 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 embedding. 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 data, the corresponding improvement is from 74.57% to 82.14%. Our experiments suggest the effectiveness of long-range sequence modeling without retraining the BERT model.