{"artifacts":[{"compression":"bzip2","distribution":"upstream-only","downloads":[{"kind":"upstream","url":"https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary/news20.binary.bz2"}],"format":"libsvm","name":"data","sha256":"2659afadb07f109afb57a7fc1ebc13b12581284fadab05b986a25c7d00690b69","size":26779006}],"description":"LIBSVM's binary classification variant of 20 Newsgroups, with 19,996 documents and 1,355,191 sparse text features. Nonzero feature vectors have approximately unit length; 42 documents have no stored features. The provider supplies a single artifact without a predefined training and test split. Feature values, binary labels, and row order are preserved.","license":{"evidence":"The LIBSVM dataset page identifies the source publication and documents normalization, but provides no license terms for this artifact. The LIBSVM software license does not establish a license for the dataset.","status":"unknown","url":"https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary.html#news20.binary"},"modality":"tabular","name":"news20-binary","preprocessing":{"datamonger":{"evidence":["spec/decoders/libsvm-v1.md"],"notes":"Datamonger decodes the provider artifact without additional statistical preprocessing.","status":"none"},"upstream":{"evidence":["https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary.html#news20.binary"],"notes":"The provider page does not describe the earlier text feature preparation or the mapping from newsgroups to the two labels. Artifact inspection finds 42 zero-feature documents; nonzero vectors have unit length within the precision of the stored values.","status":"documented","steps":[{"description":"Normalize each document's feature vector to unit length.","kind":"normalization","parameters":{"axis":"rows","norm":"l2"}}]}},"provenance":{"authors":["S. Sathiya Keerthi","Dennis DeCoste"],"citation":"S. Sathiya Keerthi and Dennis DeCoste. A Modified Finite Newton Method for Fast Solution of Large Scale Linear SVMs. Journal of Machine Learning Research, 6:341-361, 2005. https://www.jmlr.org/papers/v6/keerthi05a.html","landing_page":"https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary.html#news20.binary","provider":"LIBSVM","retrieved_at":"2026-10-20","upstream_name":"news20.binary"},"representation":{"decoder":"libsvm","decoder_version":1,"expect":{"components":[{"columns":1355191,"kind":"sparse_matrix","name":"features","rows":19996,"type":"float64"},{"kind":"vector","length":19996,"name":"response","type":"int64"}],"verification":[{"algorithm":"sha256","canonical_form":1,"digest":"9a9ef0a6d67d73e4a7d017d946e628987f28c30e365e0366a9215765a47e77cd"}]},"inputs":{"data":"data"},"options":{"duplicate_features":"error","feature_count":1355191,"index_base":1,"label_type":"int64","row_order":"source","target_name":"response"}},"schema_version":3,"source":"libsvm","tasks":[{"features":"features","name":"default","target":"response","type":"classification"}],"title":"News20 Binary","version":"1"}