license: cc-by-nc-sa-4.0
pretty_name: HEP
configs:
- config_name: default
data_files:
- split: hepth
path: Dataset/metadata-hepth.csv
- split: hepex
path: Dataset/metadata-hepex.csv
- split: astroph
path: Dataset/metadata-astroph.csv
HEP - High Energy Physics collection.
Description:
This corpus is oriented to the study of multi-labeled text classifiers. It is composed of scientific articles in the area of High Energy Physics (HEP) obtained from the CDS document server of the European Nuclear Physics Laboratory (CERN). The corpus is divided into three subsets (called partitions), where each partition is composed, in turn, of two files: one containing the records of each article (with information such as abstracts, authors and, of course, classes or keywords) in compressed XML format, and another containing a plain text version of the full article generated from the PDF available in the CERN databases (in tar + gzip format) The classes are delimited by the XML tag KEYWORD. These are the manually assigned DESY thesaurus tags. More information about the DESY thesaurus is available.
- hepth split: 18,114 Theoretical Physics documents (metadata - 5.3 Mb) (articles - 226 Mb)
- hepex split: 2,599 papers of Experimental Physics (metadata - 1.6 Mb) (articles - 28 Mb)
- astroph split: 2,716 Astrophysics documents (metadata - 1.1 Mb) (articles - 29 Mb)
Licensing Information
HEP Collection is released under the Apache-2.0 License.
Citation:
This corpus has been prepared by Arturo Montejo Ráez, with metadata provided by Jens Vigen and the help of the CDS Team.
@Article{montejo2004,
author = {Montejo-Ráez, A. and Steinberger, R. and Ureña-López, L. A.},
title = {Adaptive selection of base classifiers in one-against-all learning for large multi-labeled collections},
booktitle = {Advances in Natural Language Processing: 4th International Conference, EsTAL 2004},
pages = {1--12},
year = {2004},
editor = {Vicedo J. L. et al.},
location = {Alicante, Spain},
number = {3230},
series = {Lectures notes in artifial intelligence},
publisher = {Springer}
}