RuleBert: Teaching Soft Rules to Pre-trained Language Models

TL;DR

RuleBert introduces a method for teaching pre-trained language models to reason with soft logical rules, improving deductive reasoning and generalization to unseen rules by fine-tuning with a new dataset and loss function.

Contribution

The paper presents the first dataset and training approach for integrating soft logical rules into PLMs, enabling probabilistic reasoning and transfer of logical concepts.

Findings

High performance on reasoning tasks with unseen rules

Effective transfer of logical notions to the model

State-of-the-art results on external datasets

Abstract

While pre-trained language models (PLMs) are the go-to solution to tackle many natural language processing problems, they are still very limited in their ability to capture and to use common-sense knowledge. In fact, even if information is available in the form of approximate (soft) logical rules, it is not clear how to transfer it to a PLM in order to improve its performance for deductive reasoning tasks. Here, we aim to bridge this gap by teaching PLMs how to reason with soft Horn rules. We introduce a classification task where, given facts and soft rules, the PLM should return a prediction with a probability for a given hypothesis. We release the first dataset for this task, and we propose a revised loss function that enables the PLM to learn how to predict precise probabilities for the task. Our evaluation results show that the resulting fine-tuned models achieve very high performance, even on logical rules that were unseen at training. Moreover, we demonstrate that logical notions expressed by the rules are transferred to the fine-tuned model, yielding state-of-the-art results on external datasets.