WEPO: Web Element Preference Optimization for LLM-based Web Navigation

TL;DR

WEPO introduces an unsupervised preference learning method for LLM-based web navigation, leveraging HTML element redundancy to improve alignment with user intent, achieving state-of-the-art results on the Mind2Web benchmark.

Contribution

The paper proposes WEPO, a novel preference optimization approach that enhances LLM web navigation by using contrastive training with non-salient web elements.

Findings

WEPO outperforms previous models with a 13.8% improvement over WebAgent.

WEPO achieves a 5.3% higher score than CogAgent baseline.

Preference optimization effectively aligns LLM actions with user high-level intent.

Abstract

The rapid advancement of autonomous web navigation has significantly benefited from grounding pretrained Large Language Models (LLMs) as agents. However, current research has yet to fully leverage the redundancy of HTML elements for contrastive training. This paper introduces a novel approach to LLM-based web navigation tasks, called Web Element Preference Optimization (WEPO). WEPO utilizes unsupervised preference learning by sampling distance-based non-salient web elements as negative samples, optimizing maximum likelihood objective within Direct Preference Optimization (DPO). We evaluate WEPO on the Mind2Web benchmark and empirically demonstrate that WEPO aligns user high-level intent with output actions more effectively. The results show that our method achieved the state-of-the-art, with an improvement of 13.8% over WebAgent and 5.3% over the visual language model CogAgent baseline. Our findings underscore the potential of preference optimization to enhance web navigation and other web page based tasks, suggesting a promising direction for future research.