Saved in:
Bibliographic Details
Main Authors: Kerboua, Imene, Shayegan, Sahar Omidi, Thakkar, Megh, Lù, Xing Han, Caccia, Massimo, Eglin, Véronique, Aussem, Alexandre, Espinas, Jérémy, Lacoste, Alexandre
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2507.00210
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918076690202624
author Kerboua, Imene
Shayegan, Sahar Omidi
Thakkar, Megh
Lù, Xing Han
Caccia, Massimo
Eglin, Véronique
Aussem, Alexandre
Espinas, Jérémy
Lacoste, Alexandre
author_facet Kerboua, Imene
Shayegan, Sahar Omidi
Thakkar, Megh
Lù, Xing Han
Caccia, Massimo
Eglin, Véronique
Aussem, Alexandre
Espinas, Jérémy
Lacoste, Alexandre
contents While large language models have demonstrated impressive capabilities in web navigation tasks, the extensive context of web pages, often represented as DOM or Accessibility Tree (AxTree) structures, frequently exceeds model context limits. Current approaches like bottom-up truncation or embedding-based retrieval lose critical information about page state and action history. This is particularly problematic for adaptive planning in web agents, where understanding the current state is essential for determining future actions. We hypothesize that embedding models lack sufficient capacity to capture plan-relevant information, especially when retrieving content that supports future action prediction. This raises a fundamental question: how can retrieval methods be optimized for adaptive planning in web navigation tasks? In response, we introduce \textit{LineRetriever}, a novel approach that leverages a language model to identify and retrieve observation lines most relevant to future navigation steps. Unlike traditional retrieval methods that focus solely on semantic similarity, \textit{LineRetriever} explicitly considers the planning horizon, prioritizing elements that contribute to action prediction. Our experiments demonstrate that \textit{LineRetriever} can reduce the size of the observation at each step for the web agent while maintaining consistent performance within the context limitations.
format Preprint
id arxiv_https___arxiv_org_abs_2507_00210
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LineRetriever: Planning-Aware Observation Reduction for Web Agents
Kerboua, Imene
Shayegan, Sahar Omidi
Thakkar, Megh
Lù, Xing Han
Caccia, Massimo
Eglin, Véronique
Aussem, Alexandre
Espinas, Jérémy
Lacoste, Alexandre
Computation and Language
While large language models have demonstrated impressive capabilities in web navigation tasks, the extensive context of web pages, often represented as DOM or Accessibility Tree (AxTree) structures, frequently exceeds model context limits. Current approaches like bottom-up truncation or embedding-based retrieval lose critical information about page state and action history. This is particularly problematic for adaptive planning in web agents, where understanding the current state is essential for determining future actions. We hypothesize that embedding models lack sufficient capacity to capture plan-relevant information, especially when retrieving content that supports future action prediction. This raises a fundamental question: how can retrieval methods be optimized for adaptive planning in web navigation tasks? In response, we introduce \textit{LineRetriever}, a novel approach that leverages a language model to identify and retrieve observation lines most relevant to future navigation steps. Unlike traditional retrieval methods that focus solely on semantic similarity, \textit{LineRetriever} explicitly considers the planning horizon, prioritizing elements that contribute to action prediction. Our experiments demonstrate that \textit{LineRetriever} can reduce the size of the observation at each step for the web agent while maintaining consistent performance within the context limitations.
title LineRetriever: Planning-Aware Observation Reduction for Web Agents
topic Computation and Language
url https://arxiv.org/abs/2507.00210