# WCXB: A Multi-Type Web Content Extraction Benchmark

arXiv: https://arxiv.org/abs/2605.21097 · HTML: https://mlsanity.com/paper/2605.21097/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal RAG, one of 15 directions in Retrieval-augmented generation.
- Builds on 4 of its 20 references that are in the map.
- First posted on arXiv 20 May 2026.

## Placed in

- Retrieval-augmented generation / [Multimodal RAG](https://mlsanity.com/rag/multimodal-rag/)

## Builds on (most cited first, 4 of 4)

- [Boilerplate Removal using a Neural Sequence Labeling Model](https://mlsanity.com/paper/2004.14294/) (2020)
- [Web2Text: Deep Structured Boilerplate Removal](https://mlsanity.com/paper/1801.02607/) (2018)
- [Dripper: Token-Efficient Main HTML Extraction with a Lightweight LM](https://mlsanity.com/paper/2511.23119/) (2025)
- [Beyond a Single Extractor: Re-thinking HTML-to-Text Extraction for LLM Pretraining](https://mlsanity.com/paper/2602.19548/) (2026)

## Built on by (most cited first, 1 of 1)

- [Semantics Delivery Network: Rethinking Web Retrieval Infrastructure for LLM Agents](https://mlsanity.com/paper/2609.22486/) (2026)

## Abstract (arXiv)

Web content extraction - isolating a page's main content from surrounding boilerplate - is a prerequisite for search indexing, retrieval-augmented generation, NLP dataset construction, and large language model training. Progress in this area has been constrained by the limitations of existing evaluation benchmarks, which are small (100-800 pages), restricted to news articles, or based on web pages from over a decade ago. We introduce the Web Content Extraction Benchmark (WCXB), a dataset of 2,008 web pages from 1,613 domains spanning seven structurally distinct page types: articles, forums, products, collections, listings, documentation, and service pages. The dataset includes a 1,497-page development set and a 511-page held-out test set with matched page type distributions. Ground truth annotations were produced through a five-stage pipeline: LLM-assisted drafting, automated verification, four-pass frontier model review, snippet and quality verification scripts, and human review. We evaluate 13 extraction systems - 11 heuristic and 2 neural - and find that while top systems converge on articles (F1 = 0.93), performance diverges sharply on structured page types (F1 = 0.41-0.84), revealing blind spots invisible to existing article-only benchmarks. The dataset is released under CC-BY-4.0 with HTML source files, ground truth annotations, page type labels, and baseline results.