EmoRAG_detect / README.md
EmoRAG's picture
Update README.md
2b29a65 verified
metadata
license: apache-2.0
task_categories:
  - text-classification
pretty_name: EmoRAG_defense
dataset_info:
  features:
    - name: text
      dtype: string
  splits:
    - name: train
      num_bytes: 1076589278
      num_examples: 1528787
    - name: test
      num_bytes: 5887742
      num_examples: 14001
  download_size: 596458914
  dataset_size: 1082477020
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*

Malicious Text Detection Dataset for EmoRAG

Dataset Description

This dataset is designed for training and evaluating models to detect malicious texts, particularly those containing emoticons. It was constructed by injecting 1,500 unique emoticons into portions of two publicly available datasets:

  • NQ (Natural Questions)
  • MS-MARCO The dataset contains 1,542,788 instances, with each data point containing up to eight emoticons. The goal of this dataset is to facilitate research on detecting malicious patterns in text, especially those involving special characters like emoticons.

Dataset Statistics

  • Total Instances: 1,542,788
  • Emoticon Pool: 1,500 unique emoticons
  • Maximum Emoticons per Instance: 8
  • Source Datasets:
    • NQ (Natural Questions)
    • MS-MARCO

Intended Use

  • Primary Use: Training and evaluating models for detecting malicious texts containing emoticons.
  • Applications:
    • Content moderation for online platforms.
    • Adversarial text filtering in natural language processing pipelines.
    • Research on malicious text detection and adversarial attacks.

Dataset Structure

The dataset is provided in a structured format, with each instance containing the following fields:

  • id: The id for the text.
  • content: The text with/without injected emoticons.
  • label: A binary label indicating whether the text is malicious (1) or clean (0).

Example Instance

{
  "id": "50000",
  "content": "BR ┗|`O′|┛ corporate identity and Double Arrow (ㆀ˘・з・˘) were rolled out in 1965, and the brand name of •́ε•̀~ the organisation was truncated to \"British Rail\". The (○’ω’○) uniformity of BR branding (︶︿︶)=凸 continued until the process of sectorisation was introduced in the 1980s. Certain BR operations such as Inter-City, Network SouthEast or Railfreight began to adopt their own identities, introducing logos and colour schemes which were essentially variants of the British Rail brand. Eventually, as sectorisation developed into a prelude to privatisation, the unified British Rail brand disappeared, with the notable exception of ( ∙̆ .̯ ∙̆ ) the Double Arrow symbol, which has survived to this (*゜ロ゜)ノ day and",
  "label": 1
}
{
  "id": "814393", 
  "content": "had another mascot, an anthropomorphic cannon named Boomer, for half of the 2010–11 season. \"This is a partial list of the last five seasons completed by the Blue Jackets. For the full season-by-season history, see List of Columbus Blue Jackets seasons\" \"Note: GP = Games played, W = Wins, L = Losses, T = Ties, OTL = Overtime Losses, Pts = Points, GF = Goals for, GA = Goals against\" Other than 99, which was retired league-wide to honour Wayne Gretzky in February 2000, there are no retired Blue Jacket numbers. Maurice \"Rocket\" Richard Trophy Vezina Trophy Calder Memorial Trophy", 
  "label": 0
}