RecKG Web lets you create, integrate and visualize recommendation datasets in KG format.

Published in SoftwareX "RecKG-Web: A web-based interactive visualization tool for standardized knowledge graphs in recommender systems."

Web implementation of the paper "RecKG: Knowledge Graph for Recommender Systems". For more information about RecKG, visit the official RecKG Repository. You can download RecKG paper from arxiv.

• Users can upload recommendation datasets as separate CSV files for user data, item data, and interaction data.
  • Each file’s header should represent the attribute names to be converted into the RecKG format.
  • Example item data headers (MovieLens):
    • movie_id
    • movie_title
    • release_date
    • video_release_date
    • imdb_url
    • genre_list

• Map attributes from the uploaded files to match the RecKG format.
  • Example item data mapping:
    • movie_id → item
    • movie_title → item_name
    • video_release_date → release_date
    • type → genre_list
• After mapping, convert the dataset into the RecKG format.

{
  "item_data": {
    "item": "string",
    "item_name": "string",
    "performer": "string",
    "type": "string",
    "release_date": "string"
  },
  "user_data": {
    "user": "string",
    "age": "string",
    "gender": "string",
    "occupation": "string",
    "residence": "string"
  },
  "interaction_data": {
    "user": "string",
    "item": "string",
    "rating": "string",
    "interaction": [
      "string"
    ]
  }
}

• Integrate Two KGs

  • Functionality to merge two datasets in the RecKG format.

• Graph Visualization of Sample Data

  • Visualize the transformed dataset using D3.js.
  • Users can choose between a user-based or item-based view.
  • Options to set:
    • Initial number of nodes
    • Maximum number of interactions

• Download Processed RecKG Data

  • Provides an option to download the dataset in RecKG format after processing.