We evaluate the quality of article body extraction for commercial services Zyte Automatic Extraction (ours), Diffbot and open-source libraries newspaper3k, readability-lxml, dragnet, boilerpipe, html-text, trafilatura, go-readability, Readability.js, Go-DomDistiller. news-please. Goose3, inscriptis, html2text, jusText, BeautifulSoup. We release evaluation datasets and scripts, and provide more details in a whitepaper.

Article extraction is a task of extracting certain fields of an article (e.g. news or blog post), such as headline, article body, publication date, authors, etc. Article extraction systems must work on any web-site. Here we evaluate only the article body field, as this is one of the most important fields and one of the hardest to get right.

Results of the initial evaluation, done in November 2019:

                  version   F1             precision      recall         accuracy
AutoExtract       Nov 2019  0.970 ± 0.005  0.984 ± 0.002  0.956 ± 0.010  0.470 ± 0.037
Diffbot           Nov 2019  0.951 ± 0.010  0.958 ± 0.009  0.944 ± 0.013  0.348 ± 0.038
boilerpipe        ab3694d   0.860 ± 0.016  0.850 ± 0.016  0.870 ± 0.020  0.006 ± 0.006
dragnet           1b65e7b   0.907 ± 0.014  0.925 ± 0.013  0.889 ± 0.019  0.221 ± 0.030
html-text         0.5.1     0.665 ± 0.015  0.500 ± 0.017  0.994 ± 0.001  0.000 ± 0.000
newspaper3k       0.2.8     0.912 ± 0.014  0.917 ± 0.014  0.906 ± 0.018  0.260 ± 0.032
readability-lxml  0.7.1     0.922 ± 0.014  0.913 ± 0.014  0.931 ± 0.016  0.315 ± 0.035
xpath-text        4.4.2     0.394 ± 0.020  0.246 ± 0.016  0.992 ± 0.001  0.000 ± 0.000

Result of packages added after original evaluation:

                  version   F1             precision      recall         accuracy
trafilatura       0.5.1     0.945 ± 0.009  0.925 ± 0.011  0.966 ± 0.009  0.221 ± 0.031
go_readability    bdc8717   0.943 ± 0.007  0.912 ± 0.009  0.975 ± 0.007  0.210 ± 0.030
readability_js    Feb 2021  0.887 ± 0.012  0.853 ± 0.013  0.924 ± 0.012  0.149 ± 0.026
go_domdistiller   1c90a88   0.927 ± 0.007  0.901 ± 0.010  0.956 ± 0.010  0.066 ± 0.018
news_please       1.5.17    0.911 ± 0.014  0.917 ± 0.013  0.906 ± 0.018  0.249 ± 0.032
goose3            3.1.8     0.887 ± 0.016  0.930 ± 0.015  0.847 ± 0.021  0.227 ± 0.032
inscriptis        1.1.2     0.679 ± 0.015  0.517 ± 0.017  0.993 ± 0.001  0.000 ± 0.000
html2text         2020.1.16 0.662 ± 0.015  0.499 ± 0.017  0.983 ± 0.002  0.000 ± 0.000
justext           2.2.0     0.802 ± 0.018  0.858 ± 0.017  0.754 ± 0.028  0.088 ± 0.021
beautifulsoup     4.9.3     0.665 ± 0.015  0.499 ± 0.017  0.994 ± 0.001  0.000 ± 0.000

Below you can find more details about the packages and result reproduction.

More details are available:

• In the whitepaper at https://www.zyte.com/whitepaper-ebook/in-depth-analysis-and-evaluation-on-the-quality-of-article-body-extraction/
• In a technical report attached to the v1.0.0 release at https://github.com/scrapinghub/article-extraction-benchmark/releases/tag/v1.0.0

Clone this repo, and use Python 3.6+.

Evaluation does not require any dependencies. Dependencies listed in requirements.txt are only for re-generating output files for open-source article extraction libraries. See below for their installation details.

JSON data format: a dictionary which maps item ids to dictionaries, with the following fields:

• articleBody: text of the article
• url: page url (optional)

All files should have the same keys. Ground truth is in ground-truth.json, predictions from different systems is in output/*.json files.

HTML files are in html folder. They were fetched with Splash headless browser with JS disabled by default. They are gzip-compressed and utf-8 encoded.

Screenshots of all pages are not in the repo, they are available on github in the "Releases" section: https://github.com/scrapinghub/article-extraction-benchmark/releases

In addition to benchmarking AutoExtract and Diffbot services, we also benchmark several open-source libraries that work directly on HTML files without a need for rendering or external resources:

• newspaper3k: https://github.com/codelucas/newspaper
• readability-lxml: https://github.com/buriy/python-readability
• dragnet: https://github.com/dragnet-org/dragnet
• boilerpipe: https://github.com/misja/python-boilerpipe
• html-text: https://github.com/TeamHG-Memex/html-text - this is a baseline which extracts the full text of HTML page
• trafilatura: https://github.com/adbar/trafilatura contributed by the author at #4
• go-readability: https://github.com/go-shiori/go-readability
• Readability.js: https://github.com/mozilla/readability
• Go-DomDistiller: https://github.com/markusmobius/go-domdistiller
• news-please: https://github.com/fhamborg/news-please
• Goose3: https://github.com/goose3/goose3
• inscriptis: https://github.com/weblyzard/inscriptis - converts HTML to text with a particular emphasis on nested tables
• html2text: https://github.com/Alir3z4/html2text - converts HTML pages to Markup language
• jusText: https://github.com/miso-belica/jusText - Heuristic based boilerplate removal tool
• BeautifulSoup: https://www.crummy.com/software/BeautifulSoup/bs4/doc/ - Python library for pulling data out of HTML and XML files.

Output from these libraries is already present in the repo in output/*.json files. They were generated with extractors/run_*.py files.

All dependencies are in requirements.txt. Note that dragnet may fail to install at first try, as you need to have numpy and Cython installed, and have libxml2 headers (libxml2-dev on Ubuntu).

boilerpipe requires a custom installation: use python2, you also need Java (e.g. install default-jre in Ubuntu), install it with pip install -e git+https://github.com/misja/python-boilerpipe.git@ab3694d7bf695b73f0684a028e70aa816d63e6cb#egg=boilerpipe

go-readability requires a custom installation: see README in extractors/go_readability.

Readability.js require a custom installation: install nodejs and install cli tool: npm install -g [email protected]

Go-DomDistiller requires a custom installation: see README in extractors/go_domdistiller.

For evaluation, run:

python3 evaluate.py

We report precision, recall, F1, accuracy and their standard deviation estimated with bootstrap. Please refer to the technical report for more details.

License is MIT.