{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:5ESWMXDADW2EHHK3FGARUE4MJD","short_pith_number":"pith:5ESWMXDA","schema_version":"1.0","canonical_sha256":"e925665c601db4439d5b29811a138c48d8c5872dea8d0e3a2d08a47d3858d3c7","source":{"kind":"arxiv","id":"2211.00974","version":2},"attestation_state":"computed","paper":{"title":"Processing Long Legal Documents with Pre-trained Transformers: Modding LegalBERT and Longformer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dimitris Mamakas, Ilias Chalkidis, Ion Androutsopoulos, Petros Tsotsi","submitted_at":"2022-11-02T09:27:01Z","abstract_excerpt":"Pre-trained Transformers currently dominate most NLP tasks. They impose, however, limits on the maximum input length (512 sub-words in BERT), which are too restrictive in the legal domain. Even sparse-attention models, such as Longformer and BigBird, which increase the maximum input length to 4,096 sub-words, severely truncate texts in three of the six datasets of LexGLUE. Simpler linear classifiers with TF-IDF features can handle texts of any length, require far less resources to train and deploy, but are usually outperformed by pre-trained Transformers. We explore two directions to cope with"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2211.00974","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-11-02T09:27:01Z","cross_cats_sorted":[],"title_canon_sha256":"088a385ca7e4d166393d8fd58076381f791b10e843cf6a92194ef13ead17ae65","abstract_canon_sha256":"fc8aa51bfc7bb856d7681359c2508095d5c3d4cdbb745f06c9fb0c07dda4cacf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:15:00.768303Z","signature_b64":"78eWGYWsJtppeY+STzy92dUdMzXfEy+a7BIhtNJirQmBmRua1qaGoszfJVcoeU61N+rG5D9oSYwYxSzlHVQHBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e925665c601db4439d5b29811a138c48d8c5872dea8d0e3a2d08a47d3858d3c7","last_reissued_at":"2026-07-05T05:15:00.767856Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:15:00.767856Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Processing Long Legal Documents with Pre-trained Transformers: Modding LegalBERT and Longformer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dimitris Mamakas, Ilias Chalkidis, Ion Androutsopoulos, Petros Tsotsi","submitted_at":"2022-11-02T09:27:01Z","abstract_excerpt":"Pre-trained Transformers currently dominate most NLP tasks. They impose, however, limits on the maximum input length (512 sub-words in BERT), which are too restrictive in the legal domain. Even sparse-attention models, such as Longformer and BigBird, which increase the maximum input length to 4,096 sub-words, severely truncate texts in three of the six datasets of LexGLUE. Simpler linear classifiers with TF-IDF features can handle texts of any length, require far less resources to train and deploy, but are usually outperformed by pre-trained Transformers. We explore two directions to cope with"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2211.00974","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2211.00974/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2211.00974","created_at":"2026-07-05T05:15:00.767915+00:00"},{"alias_kind":"arxiv_version","alias_value":"2211.00974v2","created_at":"2026-07-05T05:15:00.767915+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2211.00974","created_at":"2026-07-05T05:15:00.767915+00:00"},{"alias_kind":"pith_short_12","alias_value":"5ESWMXDADW2E","created_at":"2026-07-05T05:15:00.767915+00:00"},{"alias_kind":"pith_short_16","alias_value":"5ESWMXDADW2EHHK3","created_at":"2026-07-05T05:15:00.767915+00:00"},{"alias_kind":"pith_short_8","alias_value":"5ESWMXDA","created_at":"2026-07-05T05:15:00.767915+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.01058","citing_title":"A Data Science Approach to Calcutta High Court Judgments: An Efficient LLM and RAG-powered Framework for Summarization and Similar Cases Retrieval","ref_index":8,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5ESWMXDADW2EHHK3FGARUE4MJD","json":"https://pith.science/pith/5ESWMXDADW2EHHK3FGARUE4MJD.json","graph_json":"https://pith.science/api/pith-number/5ESWMXDADW2EHHK3FGARUE4MJD/graph.json","events_json":"https://pith.science/api/pith-number/5ESWMXDADW2EHHK3FGARUE4MJD/events.json","paper":"https://pith.science/paper/5ESWMXDA"},"agent_actions":{"view_html":"https://pith.science/pith/5ESWMXDADW2EHHK3FGARUE4MJD","download_json":"https://pith.science/pith/5ESWMXDADW2EHHK3FGARUE4MJD.json","view_paper":"https://pith.science/paper/5ESWMXDA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2211.00974&json=true","fetch_graph":"https://pith.science/api/pith-number/5ESWMXDADW2EHHK3FGARUE4MJD/graph.json","fetch_events":"https://pith.science/api/pith-number/5ESWMXDADW2EHHK3FGARUE4MJD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5ESWMXDADW2EHHK3FGARUE4MJD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5ESWMXDADW2EHHK3FGARUE4MJD/action/storage_attestation","attest_author":"https://pith.science/pith/5ESWMXDADW2EHHK3FGARUE4MJD/action/author_attestation","sign_citation":"https://pith.science/pith/5ESWMXDADW2EHHK3FGARUE4MJD/action/citation_signature","submit_replication":"https://pith.science/pith/5ESWMXDADW2EHHK3FGARUE4MJD/action/replication_record"}},"created_at":"2026-07-05T05:15:00.767915+00:00","updated_at":"2026-07-05T05:15:00.767915+00:00"}