{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "sennrich-haddow-birch-subword-nmt-2015",
    "title": "Character-merge subword units for open-vocabulary neural translation",
    "summary": "Adapted byte-pair compression ideas to learn character-sequence merges within words, representing rare and unseen words as reusable subword sequences in neural translation.",
    "boundary": "The adaptation merges characters or character sequences, not necessarily raw bytes. It inherits Gage's compression algorithm and earlier subword ideas. A finite alphabet and learned representation do not guarantee correct translation of every unseen string.",
    "area": "Tokenization for neural translation",
    "topic": "Artificial intelligence",
    "year": 2015,
    "lane": "HUMAN",
    "result": "ALGORITHM",
    "package_id": "sennrich-haddow-birch-subword-nmt-2015",
    "target": {
      "record_id": "urn:uuid:f509a458-5a77-4671-a25c-82ee418b0be1",
      "version": 1,
      "content_hash": "sha256:52fda9cc18417c069bd31fb3009073eecd5f9ae5a2ff8b8792c62e04924aa1dc"
    },
    "humans": [
      "Rico Sennrich",
      "Barry Haddow",
      "Alexandra Birch"
    ],
    "authors_complete": true,
    "human_role_note": "Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.",
    "agents": [],
    "related": [],
    "subresults": [],
    "scientific_assessment": {
      "source_review": "SOURCE_CURATED",
      "scientific_review": "NOT_REVIEWED_BY_WRS",
      "formal_replay": "NOT_RUN",
      "provider_attestation": "NONE"
    },
    "sources": [
      {
        "id": "cur20261015-arxiv-1508-07909v1",
        "title": "Neural Machine Translation of Rare Words with Subword Units",
        "url": "https://arxiv.org/abs/1508.07909v1"
      }
    ],
    "points": "175000",
    "published_grade": {
      "id": "sennrich-haddow-birch-subword-nmt-2015",
      "entry_id": "sennrich-haddow-birch-subword-nmt-2015",
      "band": "A",
      "modifier": "+",
      "target": {
        "record_id": "urn:uuid:f509a458-5a77-4671-a25c-82ee418b0be1",
        "version": 1,
        "content_hash": "sha256:52fda9cc18417c069bd31fb3009073eecd5f9ae5a2ff8b8792c62e04924aa1dc"
      },
      "what_changed": "Adapted byte-pair compression ideas to learn character-sequence merges within words, representing rare and unseen words as reusable subword sequences in neural translation.",
      "scope": "Adapted byte-pair compression ideas to learn character-sequence merges within words, representing rare and unseen words as reusable subword sequences in neural translation. The adaptation merges characters or character sequences, not necessarily raw bytes. It inherits Gage's compression algorithm and earlier subword ideas. A finite alphabet and learned representation do not guarantee correct translation of every unseen string.",
      "status": "PROVISIONAL_EDITORIAL",
      "scientific_correctness_certified": false
    },
    "assessment_disposition": {
      "status": "ASSESSED",
      "label": "Provisionally assessed",
      "reason": "A current grade and separate reference score identify this exact contribution.",
      "target": {
        "record_id": "urn:uuid:f509a458-5a77-4671-a25c-82ee418b0be1",
        "version": 1,
        "content_hash": "sha256:52fda9cc18417c069bd31fb3009073eecd5f9ae5a2ff8b8792c62e04924aa1dc"
      },
      "edition": "ratings-e039106f30dec177"
    }
  },
  "version": {
    "kind": "PUBLIC_RECORD_VIEW",
    "previous": null,
    "published_at": "2026-10-16T00:00:00Z",
    "publisher": {
      "principal_id": "urn:uuid:80116931-785b-40a2-a24b-9d59db0c7e33",
      "role": "WRS_EDITORIAL_PUBLICATION"
    },
    "record": {
      "contributors": [
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Rico Sennrich",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Neural Machine Translation of Rare Words with Subword Units",
            "url": "https://arxiv.org/abs/1508.07909v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Barry Haddow",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Neural Machine Translation of Rare Words with Subword Units",
            "url": "https://arxiv.org/abs/1508.07909v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Alexandra Birch",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Neural Machine Translation of Rare Words with Subword Units",
            "url": "https://arxiv.org/abs/1508.07909v1"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "Neural Machine Translation of Rare Words with Subword Units",
            "url": "https://arxiv.org/abs/1508.07909v1"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "The adaptation merges characters or character sequences, not necessarily raw bytes. It inherits Gage's compression algorithm and earlier subword ideas. A finite alphabet and learned representation do not guarantee correct translation of every unseen string.",
        "domain": "Tokenization for neural translation"
      },
      "sources": [
        {
          "citation": "Neural Machine Translation of Rare Words with Subword Units",
          "url": "https://arxiv.org/abs/1508.07909v1"
        }
      ],
      "statement": "# Character-merge subword units for open-vocabulary neural translation\n\nAdapted byte-pair compression ideas to learn character-sequence merges within words, representing rare and unseen words as reusable subword sequences in neural translation.\n\n## Scientific scope\n\nThe adaptation merges characters or character sequences, not necessarily raw bytes. It inherits Gage's compression algorithm and earlier subword ideas. A finite alphabet and learned representation do not guarantee correct translation of every unseen string.\n\n## Reading status\n\nThis is a source-linked WRS catalogue note. A full explanatory guide and independent scientific reconstruction have not been completed for this entry. Read the identified original work for its complete narrative and arguments.\n\n## People and systems\n\nRico Sennrich; Barry Haddow; Alexandra Birch\n\nSource-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.\n\n\n\n## Original sources\n\n- [Neural Machine Translation of Rare Words with Subword Units](https://arxiv.org/abs/1508.07909v1)\n",
      "summary": "Adapted byte-pair compression ideas to learn character-sequence merges within words, representing rare and unseen words as reusable subword sequences in neural translation.",
      "title": "Character-merge subword units for open-vocabulary neural translation"
    },
    "record_id": "urn:uuid:f509a458-5a77-4671-a25c-82ee418b0be1",
    "version": 1,
    "source_ref": {
      "record_id": "urn:uuid:f509a458-5a77-4671-a25c-82ee418b0be1",
      "version": 1,
      "content_hash": "sha256:52fda9cc18417c069bd31fb3009073eecd5f9ae5a2ff8b8792c62e04924aa1dc"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:4760a8b31e2be7d96c4bd4ff82c561c6dd625e5714e1fa7b02f5e024c55c38c6"
  },
  "target": {
    "record_id": "urn:uuid:f509a458-5a77-4671-a25c-82ee418b0be1",
    "version": 1,
    "content_hash": "sha256:52fda9cc18417c069bd31fb3009073eecd5f9ae5a2ff8b8792c62e04924aa1dc"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "sennrich-haddow-birch-subword-nmt-2015",
    "entry_id": "sennrich-haddow-birch-subword-nmt-2015",
    "revision": 1,
    "title": "Character-merge subword units for open-vocabulary neural translation",
    "context": "Tokenization for neural translation",
    "track": "Human-designed research method or resource",
    "band": "A",
    "modifier": "+",
    "status": "PROVISIONAL_EDITORIAL",
    "assessed_on": "2026-10-16",
    "evidence_cutoff": "2026-10-16",
    "achievement_state": "SUPPORTED_SCOPED_RESULT",
    "what_changed": "Adapted byte-pair compression ideas to learn character-sequence merges within words, representing rare and unseen words as reusable subword sequences in neural translation.",
    "scope": "Adapted byte-pair compression ideas to learn character-sequence merges within words, representing rare and unseen words as reusable subword sequences in neural translation. The adaptation merges characters or character sequences, not necessarily raw bytes. It inherits Gage's compression algorithm and earlier subword ideas. A finite alphabet and learned representation do not guarantee correct translation of every unseen string.",
    "evidence_summary": "The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. The adaptation merges characters or character sequences, not necessarily raw bytes. It inherits Gage's compression algorithm and earlier subword ideas. A finite alphabet and learned representation do not guarantee correct translation of every unseen string.",
    "references": [
      {
        "title": "Neural Machine Translation of Rare Words with Subword Units",
        "url": "https://arxiv.org/abs/1508.07909v1",
        "locator": "Original contribution and stated assumptions, evaluation scope and limitations; source-linked assessment, not independent execution or proof certification."
      }
    ],
    "roles": [
      {
        "who": "Source-listed authors and technical contributors",
        "credit": "SCIENCE",
        "did": "Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities."
      }
    ],
    "target": {
      "record_id": "urn:uuid:f509a458-5a77-4671-a25c-82ee418b0be1",
      "version": 1,
      "content_hash": "sha256:52fda9cc18417c069bd31fb3009073eecd5f9ae5a2ff8b8792c62e04924aa1dc"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "175000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": null,
  "reading_status": "SOURCE_LINKED_NOTE"
}