A dataset used in the paper "Evaluating the Helpfulness of Linked Entities to Readers"
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
iitb_dataset
README.md
annotations.json
splitter.py

README.md

el-helpfulness-dataset

This dataset is created for the paper Evaluating the Helpfulness of Linked Entities to Readers.

The paper aims to automatically detect entities that are helpful to document readers. The dataset is based on an entity linking dataset called IITB EL dataset.

The dataset is contained in annotations.json. Each annotation contains the following properties:

  • annotator: The annotator's ID
  • label: The annotated label that represents whether the entity is helpful to readers
  • wikipediaTitle: The Wikipedia title of the entity
  • iitbDocName: The file name of the document in the IITB dataset
  • iitbLength: The entity length
  • iitbLength: The entity offset
  • iitbUserId: The user ID in the IITB dataset

A label can be one of the following:

  • Very helpful (YY): Converting this keyword into a link is highly helpful to readers.
  • Helpful (Y): Converting this keyword into a link might be helpful to some readers.
  • Rarely helpful (N): Converting this keyword into a link is rarely helpful to readers.
  • Not helpful (NN): Converting this keyword into a link is not helpful to readers.

If you use this dataset in your research, please cite the following paper:

@inproceedings{LINKIFY2014,
 author = {Yamada, Ikuya and Ito, Tomotaka and Usami, Shinnosuke and Takagi, Shinsuke and Takeda, Hideaki and Takefuji, Yoshiyasu},
 title = {Evaluating the Helpfulness of Linked Entities to Readers},
 booktitle = {Proceedings of the 25th ACM Conference on Hypertext and Social Media},
 location = {Santiago, Chile},
 year = {2014},
 pages = {169--178}
}

License

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.