{"artifacts":[{"compression":"bzip2","distribution":"upstream-only","downloads":[{"kind":"upstream","url":"https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary/rcv1_train.binary.bz2"}],"format":"libsvm","name":"train","sha256":"dda0c6381811484e284e70dd0641335917debce50948346416fde7fb32fad2e2","size":13730096},{"compression":"bzip2","distribution":"upstream-only","downloads":[{"kind":"upstream","url":"https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary/rcv1_test.binary.bz2"}],"format":"libsvm","name":"test","sha256":"975ab6c4fd8ea2a88b82e7ba2659ef21bf980ac61b3d8f13d6991739ec0f7aa7","size":453311155}],"description":"LIBSVM's binary classification variant of Reuters Corpus Volume 1, with 47,236 sparse text features and predefined training and test sets. Positive labels combine CCAT and ECAT, and negative labels combine GCAT and MCAT. Documents assigned to both groups are removed. The provider's feature values, labels, splits, and row order are preserved.","license":{"evidence":"The LIBSVM dataset page documents the source and binary label mapping but provides no license terms for these artifacts. NIST identifies Reuters copyright and organizational and individual agreements for the source corpus. No license or redistribution grant has been established for the LIBSVM transformation.","status":"unknown","url":"https://trec.nist.gov/data/reuters/reuters.html"},"modality":"tabular","name":"rcv1-binary","preprocessing":{"datamonger":{"evidence":["spec/decoders/libsvm-v1.md"],"notes":"Datamonger decodes the provider artifacts without additional statistical preprocessing.","status":"none"},"upstream":{"evidence":["https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary.html#rcv1.binary"],"notes":"The provider feature values, labels, splits, and row order are preserved. Earlier text feature preparation is not described here.","status":"documented","steps":[{"description":"Combine CCAT and ECAT into the positive class and GCAT and MCAT into the negative class.","kind":"response-dichotomization","parameters":{"negative_classes":"GCAT, MCAT","positive_classes":"CCAT, ECAT"}},{"description":"Remove documents assigned to both the positive and negative groups.","kind":"sample-selection","parameters":{"predicate":"retain documents assigned to only one label group"}}]}},"provenance":{"authors":["David D. Lewis","Yiming Yang","Tony G. Rose","Fan Li"],"citation":"David D. Lewis, Yiming Yang, Tony G. Rose, and Fan Li. RCV1: A New Benchmark Collection for Text Categorization Research. Journal of Machine Learning Research, 5:361-397, 2004. https://www.jmlr.org/papers/v5/lewis04a.html","landing_page":"https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary.html#rcv1.binary","provider":"LIBSVM","retrieved_at":"2026-10-19","upstream_name":"rcv1.binary"},"representation":{"decoder":"libsvm-split","decoder_version":1,"expect":{"components":[{"columns":47236,"kind":"sparse_matrix","name":"train_features","rows":20242,"type":"float64"},{"kind":"vector","length":20242,"name":"train_response","type":"int64"},{"columns":47236,"kind":"sparse_matrix","name":"test_features","rows":677399,"type":"float64"},{"kind":"vector","length":677399,"name":"test_response","type":"int64"}],"verification":[{"algorithm":"sha256","canonical_form":1,"digest":"7bffc1f359c7bc2cc7d2c2eb54965f24f3f6e67998566a4efa792fc6c6629934"}]},"inputs":{"test":"test","train":"train"},"options":{"duplicate_features":"error","feature_count":47236,"index_base":1,"label_type":"int64","row_order":"source","target_name":"response"}},"schema_version":3,"source":"libsvm","tasks":[{"name":"default","splits":{"test":{"features":"test_features","target":"test_response"},"train":{"features":"train_features","target":"train_response"}},"type":"classification"}],"title":"RCV1 Binary","version":"1"}