How to Encode Domain Information in Relation Classification

TL;DR

This paper investigates encoding domain information in relation classification models to improve multi-domain performance, showing that domain-aware models significantly outperform baselines, especially for domain-specific relations.

Contribution

It introduces a method to encode domain information in relation classification models, enhancing multi-domain performance and revealing how different relation types benefit variably.

Findings

Models improve > 2 Macro-F1 over baselines.

Domain-dependent relations benefit most from encoding.

Similar-space relations benefit least from encoding.

Abstract

Current language models require a lot of training data to obtain high performance. For Relation Classification (RC), many datasets are domain-specific, so combining datasets to obtain better performance is non-trivial. We explore a multi-domain training setup for RC, and attempt to improve performance by encoding domain information. Our proposed models improve > 2 Macro-F1 against the baseline setup, and our analysis reveals that not all the labels benefit the same: The classes which occupy a similar space across domains (i.e., their interpretation is close across them, for example "physical") benefit the least, while domain-dependent relations (e.g., "part-of'') improve the most when encoding domain information.