gerbejon's picture
Upload dataset
a7085fe verified
|
raw
history blame
3.05 kB
metadata
dataset_info:
  features:
    - name: hyu
      dtype: string
    - name: xpath
      dtype: string
    - name: page_id
      dtype: string
    - name: annotator_id
      dtype: string
    - name: node_parents_labels
      sequence: string
    - name: tag
      dtype: string
    - name: node_label
      sequence: string
    - name: node_children_labels
      sequence: string
  splits:
    - name: train
      num_bytes: 341079832
      num_examples: 2080005
    - name: test
      num_bytes: 341079832
      num_examples: 2080005
  download_size: 71823258
  dataset_size: 682159664
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*

WebClasSeg25 Dataset Collection

The WebClasSeg25 dataset series provides structured data for webpage segment detection, offering both functional segmentation (FC) and maturity classification (MC). The datasets are designed for accessibility and ease of use in model training, with different levels of granularity.

Dataset Collection Overview

WebClasSeg25-html

  • Contains the full HTML code of each website.
  • Each page is assigned a unique key, which links it to the segmentation datasets.

WebClasSeg25-visual-fc

  • Provides visual segmentations based on functional classification.
  • Each segmented page is linked to its corresponding HTML file using a shared key.

WebClasSeg25-visual-mc

  • Provides visual segmentations based on maturity classification.
  • Each segmented page is linked to its corresponding HTML file using a shared key.

WebClasSeg25-html-nodes-mc

  • Contains individual HTML nodes for each page, labeled according to maturity classification.
  • Includes parent, node, and children classifications.
  • Provides a unique key per page and per node.

WebClasSeg25-html-nodes-fc

  • Contains individual HTML nodes for each page, labeled according to functional classification.
  • Includes parent, node, and children classifications.
  • Provides a unique key per page and per node.

Dataset Relationships

  • The WebClasSeg25-html dataset serves as the foundation, containing the full HTML structure.
  • The visual segmentation datasets (FC & MC) offer page-level segmentation linked to the HTML dataset via keys.
  • The node-level datasets (FC & MC) break down HTML into individual elements, assigning functional or maturity classifications while maintaining linkage to their parent and children nodes.

This dataset collection is structured to support a wide range of segmentation tasks, making it easier to train and evaluate models across different levels of webpage structure.