/LiterallyWikidata

LiterallyWikidata

  • The dataset and implementation described in the paper accepted in ISWC 2021.
  • A Benchmark for Knowledge Graph Completion using Numeric and Text Literals extracted from Wikidata and Wikipedia.
  • It can be used to evaluate both unimodal and multimodal Knowledge Graph Embedding approaches.
  • The collection contains three differnet datasets LitWD1K, LitWD19K, and LitWD48K.

Dataset Statstics and Analysis

LitWD1K LitWD19K LitWD48K
#Entities 1,533 18,986 47,998
#Relations 47 182 257
#Attributes 81 151 297
#Structured Triples 29,017 288,933 336,745
#Numerical Attributive Triples 10,988 63,951 324,418
#Train 26,115 260,039 303,117
#Test 1,451 14,447 16,838
#Valid 1,451 14,447 16,838

Literals

  • We have extracted labels, aliases, and descriptions from wikidata for entities, relations, and attributes. There also long text descritions for entities extracted from the summary sections of thier corresponding English, German, Russian, and Chinese Wikipedia pages.

Benchmarking

  • Link predection experiments are conducted with three models DistMult, ComplEx, and DistMultLiteral on all datasets using Pykeen.
  • the configurations for each of the models are given in the Benchmarking diectory.

##Citation

@inproceedings{GeseseAS21,
  author       = {Genet Asefa Gesese and
                  Mehwish Alam and
                  Harald Sack},
  editor       = {Andreas Hotho and
                  Eva Blomqvist and
                  Stefan Dietze and
                  Achille Fokoue and
                  Ying Ding and
                  Payam M. Barnaghi and
                  Armin Haller and
                  Mauro Dragoni and
                  Harith Alani},
  title        = {LiterallyWikidata - {A} Benchmark for Knowledge Graph Completion Using
                  Literals},
  booktitle    = {The Semantic Web - {ISWC} 2021 - 20th International Semantic Web Conference,
                  {ISWC} 2021, Virtual Event, October 24-28, 2021, Proceedings},
  series       = {Lecture Notes in Computer Science},
  volume       = {12922},
  pages        = {511--527},
  publisher    = {Springer},
  year         = {2021},
  url          = {https://doi.org/10.1007/978-3-030-88361-4\_30},
  doi          = {10.1007/978-3-030-88361-4\_30},
}

##Contact

In case of any questions please open Github issue or alternatively contact Genet Asefa Gesese or Mehwish Alam.