# Cross-Domain Web Information Extraction at Pinterest

arXiv: https://arxiv.org/abs/2508.01096 · HTML: https://mlsanity.com/paper/2508.01096/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 13 references that are in the map.
- First posted on arXiv 1 Aug 2025.

## Builds on (most cited first, 8 of 8)

- [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://mlsanity.com/paper/2204.08387/) (2022)
- [DocFormerv2: Local Features for Document Understanding](https://mlsanity.com/paper/2306.01733/) (2023)
- [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://mlsanity.com/paper/2110.08518/) (2021)
- [TransAct: Transformer-based Realtime User Action Model for Recommendation at Pinterest](https://mlsanity.com/paper/2306.00248/) (2023)
- [ItemSage: Learning Product Embeddings for Shopping Recommendations at Pinterest](https://mlsanity.com/paper/2205.11728/) (2022)
- [Simplified DOM Trees for Transferable Attribute Extraction from the Web](https://mlsanity.com/paper/2101.02415/) (2021)
- [OmniSearchSage: Multi-Task Multi-Entity Embeddings for Pinterest Search](https://mlsanity.com/paper/2404.16260/) (2024)
- [Inscriptis -- A Python-based HTML to text conversion library optimized for knowledge extraction from the Web](https://mlsanity.com/paper/2108.01454/) (2021)

## Abstract (arXiv)

The internet offers a massive repository of unstructured information, but it's a significant challenge to convert this into a structured format. At Pinterest, the ability to accurately extract structured product data from e-commerce websites is essential to enhance user experiences and improve content distribution. In this paper, we present Pinterest's system for attribute extraction, which achieves remarkable accuracy and scalability at a manageable cost. Our approach leverages a novel webpage representation that combines structural, visual, and text modalities into a compact form, optimizing it for small model learning. This representation captures each visible HTML node with its text, style and layout information. We show how this allows simple models such as eXtreme Gradient Boosting (XGBoost) to extract attributes more accurately than much more complex Large Language Models (LLMs) such as Generative Pre-trained Transformer (GPT). Our results demonstrate a system that is highly scalable, processing over 1,000 URLs per second, while being 1000 times more cost-effective than the cheapest GPT alternatives.