Bi-Link: Bridging Inductive Link Predictions from Text via Contrastive Learning of Transformers and Prompts
Peng, Bohua, Liang, Shihao, Islam, Mobarakol
–arXiv.org Artificial Intelligence
Inductive knowledge graph completion requires models to comprehend the underlying semantics and logic patterns of relations. With the advance of pretrained language models, recent research have designed transformers for link prediction tasks. However, empirical studies show that linearizing triples affects the learning of relational patterns, such as inversion and symmetry. In this paper, we propose Bi-Link, a contrastive learning framework with probabilistic syntax prompts for link predictions. Using grammatical knowledge of BERT, we efficiently search for relational prompts according to learnt syntactical patterns that generalize to large knowledge graphs. To better express symmetric relations, we design a symmetric link prediction model, establishing bidirectional linking between forward prediction and backward prediction. This bidirectional linking accommodates flexible self-ensemble strategies at test time. In our experiments, Bi-Link outperforms recent baselines on link prediction datasets (WN18RR, FB15K-237, and Wikidata5M). Furthermore, we construct Zeshel-Ind as an in-domain inductive entity linking the environment to evaluate Bi-Link. The experimental results demonstrate that our method yields robust representations which can generalize under domain shift.
arXiv.org Artificial Intelligence
Oct-26-2022
- Country:
- North America
- Dominican Republic (0.04)
- United States
- Oregon (0.04)
- Texas > Travis County
- Austin (0.05)
- Europe
- Western Europe (0.04)
- United Kingdom
- Wales (0.04)
- Scotland (0.04)
- Northern Ireland (0.04)
- England > Greater London
- London > City of Westminster (0.04)
- France > Île-de-France
- Asia > China
- North America
- Genre:
- Research Report > New Finding (0.68)
- Technology:
- Information Technology
- Information Management > Search (1.00)
- Data Science > Data Mining (1.00)
- Artificial Intelligence
- Representation & Reasoning (1.00)
- Machine Learning (1.00)
- Natural Language > Text Processing (0.46)
- Information Technology