Visual-Aware Representation of Web Pages for Machine Learning Applications

This repository contains a sample data set that demonstrates the use of web pages as a data source for visual-aware machine learning applications using the FitLayout framework.

The dataset captures the rendered pages from the imaginary bookstore available at https://books.toscrape.com/. For each book page in the book store, the data set contains two FitLayout artifacts:

• A Page that directly describes the rendered page at the box level.
• An AreaTree that provides abstraction over the rendered page in the form of a tree of visual areas, where significant areas (e.g., book title and price) are annotated with the corresponding tags.

The artifacts have been exported from the FitLayout RDF repository in the N-QUADS format, which allows easy importing them to another repository.

Contained files

• book_urls.txt -- the source URLs of the rendered pages.
• books_artifacts.zip -- the RDF graph describing all the artifacts serialized in N-QUADS format.