DISCIE -- Discriminative Closed Information Extraction

TL;DR

This paper presents DISCIE, a discriminative method for closed information extraction that leverages type and entity-specific info to outperform generative models, especially in large-scale, long-tail relation scenarios.

Contribution

The paper introduces a discriminative approach that effectively incorporates type information, achieving high accuracy with smaller models in large-scale closed information extraction.

Findings

Outperforms state-of-the-art generative models in relation extraction accuracy.

Efficiently handles large-scale datasets with millions of entities.

Type information significantly boosts extraction performance.

Abstract

This paper introduces a novel method for closed information extraction. The method employs a discriminative approach that incorporates type and entity-specific information to improve relation extraction accuracy, particularly benefiting long-tail relations. Notably, this method demonstrates superior performance compared to state-of-the-art end-to-end generative models. This is especially evident for the problem of large-scale closed information extraction where we are confronted with millions of entities and hundreds of relations. Furthermore, we emphasize the efficiency aspect by leveraging smaller models. In particular, the integration of type-information proves instrumental in achieving performance levels on par with or surpassing those of a larger generative model. This advancement holds promise for more accurate and efficient information extraction techniques.