WebShaper: Agentically Data Synthesizing via Information-Seeking Formalization

TL;DR

WebShaper introduces a formalization-driven framework for synthesizing high-quality web-based information-seeking data, enabling the training of more effective LLM-powered agents for complex tasks.

Contribution

It proposes a novel formalization method using set theory and Knowledge Projections to systematically generate training data for IS agents.

Findings

Achieves state-of-the-art performance on GAIA benchmark.

Outperforms existing open-sourced IS agents.

Demonstrates effective control over reasoning structures.

Abstract

The advent of Large Language Model (LLM)-powered agents has revolutionized artificial intelligence by enabling solutions to complex, open-ended tasks through web-based information-seeking (IS) capabilities. The scarcity of high-quality training data has limited the development of IS agents. Existing approaches typically adopt an information-driven paradigm that first collects web data and then generates questions based on the retrieval. However, this may lead to inconsistency between information structure and reasoning structure, question and answer. To mitigate, we propose a formalization-driven IS data synthesis framework WebShaper to construct a dataset. WebShaper systematically formalizes IS tasks through set theory. Central to the formalization is the concept of Knowledge Projections (KP), which enables precise control over reasoning structure by KP operation compositions. During synthesis, we begin by creating seed tasks, then use a multi-step expansion process. At each step, an agentic Expander expands the current formal question more complex with retrieval and validation tools based on our formalization. We train our model on the synthesized dataset. Experiment results demonstrate that WebShaper achieves state-of-the-art performance among open-sourced IS agents on GAIA and WebWalkerQA benchmarks.