Longitudinal Web Archive Reconstruction, Stylesheet Decay & Historical DOM Preservation (1996–2026)
Prepared for digital preservation researchers, web-archiving librarians, computational social scientists, and search infrastructure engineers. This open-access repository documents empirical stylesheet decay rates (HTTP 404 CSS/JS attrition), legacy character encoding drift (windows-1251, koi8-r, shift_jis → UTF-8), third-party tracker detoxification, and automated DOM layout recovery across 16 CMS cohorts in the Internet Archive Wayback Machine.
Unreconstructed Wayback captures with missing cache-busted CSS
Post-recovery layout & stylesheet integrity (Engine v38)
Legacy analytics, ad beacons, counters & archive toolbars
Free for university syllabi, digital libraries & NLP corpora
1. Open-Access Technical Reports & Working Papers (PDF)
Longitudinal Web Archive Reconstruction & Stylesheet Decay
Empirical study of cache-busted CSS/JS rot across WordPress, OpenCart, Joomla, Bitrix, and Drupal captures, and CDN-backed DOM reconstruction.
Legacy Character Encoding Normalization & Tracker Detox
Deterministic byte-stream transcoding for windows-1251, koi8-r, and shift_jis archives and AST removal of 25+ historical tracker classes.
Multi-Region CDX Playback & Link-Graph Continuity
Sparkline HTTP status-map analysis for anti-parking snapshot selection and standalone static packaging (.htaccess, Nginx, Edge Workers).
2. Empirical CMS Archival Survival & DOM Reconstruction Benchmark (1999–2025)
| # | CMS / Architecture Cohort | Era | Charset | Raw CSS 404 | Reconstructed CSS | Trackers Stripped | DOM Fidelity | Module |
|---|---|---|---|---|---|---|---|---|
| 01 | WordPress + Astra + Elementor | 2021-2025 | UTF-8 | 64% | 99% | 14 | 98% | Inspect |
| 02 | WordPress + WooCommerce Store | 2020-2025 | UTF-8 | 58% | 98% | 19 | 97% | Inspect |
| 03 | WordPress + ColorMag Magazine | 2019-2025 | UTF-8 | 71% | 99% | 16 | 99% | Inspect |
| 04 | OpenCart + UniShop2 + BusCache | 2022-2025 | UTF-8 | 52% | 99% | 11 | 97% | Inspect |
| 05 | Legacy HTML4 Frameset Portal | 2001-2012 | windows-1251 | 82% | 96% | 9 | 95% | Inspect |
| 06 | 1C-Bitrix Corporate & Catalog | 2014-2024 | windows-1251 | 49% | 97% | 18 | 96% | Inspect |
| 07 | Joomla 3.x / K2 Academic Portal | 2013-2022 | UTF-8 | 61% | 97% | 12 | 96% | Inspect |
| 08 | Drupal 7 / 9 Institutional Site | 2012-2023 | UTF-8 | 55% | 98% | 10 | 97% | Inspect |
| 09 | DataLife Engine (DLE) Media Portal | 2010-2021 | windows-1251 | 67% | 96% | 22 | 95% | Inspect |
| 10 | MODX Revolution Regional News | 2015-2024 | UTF-8 | 44% | 98% | 15 | 98% | Inspect |
| 11 | WordPress + GeneratePress Blog | 2018-2025 | UTF-8 | 69% | 99% | 13 | 99% | Inspect |
| 12 | WordPress + OceanWP E-Commerce | 2019-2024 | UTF-8 | 63% | 98% | 17 | 97% | Inspect |
| 13 | Static Bootstrap 4/5 Landing Hub | 2017-2025 | UTF-8 | 38% | 99% | 8 | 99% | Inspect |
| 14 | Legacy KOI8-R Academic Archive | 1999-2009 | koi8-r | 77% | 95% | 6 | 94% | Inspect |
| 15 | Shift_JIS / EUC-JP Regional Portal | 2004-2018 | shift_jis | 73% | 96% | 11 | 95% | Inspect |
| 16 | PrestaShop 1.7 Multi-Category Store | 2016-2024 | UTF-8 | 59% | 97% | 20 | 96% | Inspect |
Load Directly in Python, pandas & Hugging Face
All benchmark tables are served with CORS headers enabled and zero authentication required for reproducible research pipelines:
import pandas as pd
# Load UTF-8 Benchmark CSV directly
df = pd.read_csv("https://wayforge.org/api/research/dataset.csv")
print(df[["cms_family", "raw_css_404_rate", "dom_fidelity_score"]])
# Or load JSONL stream in Hugging Face datasets
from datasets import load_dataset
ds = load_dataset("json", data_files="https://wayforge.org/api/research/dataset.jsonl")Official Citation (APA 7th & BibTeX)
APA 7th:WayForge Digital Web Preservation & Archival Reconstruction Lab. (2026). Longitudinal web archive reconstruction, stylesheet decay, and historical DOM preservation across 1996–2026 CDX captures (Technical Report WF-TR-2026-01). WayForge Research. https://wayforge.org/research
@techreport{wayforge2026reconstruction,
title = {Longitudinal Web Archive Reconstruction, Stylesheet Decay, and Historical DOM Preservation Across 1996--2026 CDX Captures},
author = {{WayForge Digital Web Preservation \& Archival Reconstruction Lab}},
institution = {WayForge Research Lab},
year = {2026},
month = oct,
number = {WF-TR-2026-01},
url = {https://wayforge.org/research},
note = {Open-Access Benchmark Dataset and Automated DOM Reconstruction Engine (CC BY 4.0). Available at https://wayforge.org/research}
}
@misc{wayforge2026dataset,
title = {WayForge 2026 Historical Web Preservation \& DOM Reconstruction Benchmark Dataset (1996--2026)},
author = {{WayForge Digital Web Preservation \& Archival Reconstruction Lab}},
year = {2026},
howpublished = {\url{https://wayforge.org/api/research/dataset.csv}},
note = {Creative Commons Attribution 4.0 International (CC BY 4.0). Documentation: https://wayforge.org/docs}
}
3. Permanent Institutional Repositories, DataCite DOIs & Academic Mirrors
The WayForge 2026 Longitudinal Web Archive Reconstruction Corpus and Working Papers are permanently deposited across university, governmental, and international scientific repositories: