Welcome to FranKGraphbench’s documentation!

This project is an outcome of Google Summer of Code 2023 project at DBpedia, named FranKGraphBench: Knowledge Graph Aware Recommender Systems Framework for Benchmarking.

This project have two main components:

• Data Integration: for enriching standard Recommender Systems datasets using DBpedia.

  • As part of the Data Integration module we fixed and used the DBpedia 2016-10 core version, accessible through the endpoint https://dbfk25.aksw.org/sparql/ and with .ttl files downloadable https://dbfk25.aksw.org/.

• Framework: for running reproducible experiments on enriched and non-enriched datasets through a .yml configuration file.

Getting Started:

• Introduction
• Quick Start
  • Installation and usage
• Supported Datasets, Models and Others
  • Datasets
  • Models
  • Pre-processing Methods
  • Splitting Methods
  • Evaluation Metrics
  • Chart Generation
• Setting up configuration file
  • dataset
  • preprocess
  • split
  • models
  • evaluation
  • report

Contributing:

• Introduction
  • Basic Workflow
  • More details
• Add Dataset
• Add Recommender System Model
• Add Evaluation Metric
• Add Splitting Method
• Add Pre-processing method
• Improve matching with DBpedia
  • Guide for wrongly matched/unmatched items
• Bug Fixing

Benchmark Results:

• Overall Results
  • ml-100k
  • ml-100k_enriched
  • ml-1m
  • ml-1m_enriched
  • lastfm
  • lastfm_enriched
  • douban-movie
  • douban-movie_enriched
  • mind-small
  • mind-small_enriched
• Embedding Dimension Ablation
  • ml-100k kge_embedding
  • ml-100k gnn_embedding

Indices and tables

• Index

• Module Index

• Search Page