{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "zhao-et-al-recd-2022",
    "title": "RecD: session-aware feature deduplication across recommendation training",
    "summary": "Coalesced session samples across logging, ETL and training, introducing inverse-keyed jagged tensors that share repeated feature values while retaining inverse mappings to their training examples.",
    "boundary": "Deduplicates repeated features, not independent labels or arbitrary examples. Session grouping changes ordering and requires pipeline support; efficiency depends on actual session reuse and feature distributions. Reported storage, preprocessing and training gains are different metrics on selected industrial workloads.",
    "area": "Recommendation training infrastructure",
    "topic": "Artificial intelligence",
    "year": 2022,
    "lane": "HUMAN",
    "result": "ALGORITHM",
    "package_id": "zhao-et-al-recd-2022",
    "target": {
      "record_id": "urn:uuid:eeb75c07-1597-4107-a466-2d1aca5997de",
      "version": 1,
      "content_hash": "sha256:54d01826b02f02fae945896ee878020712616ce11a00ab17370e99cad887590d"
    },
    "humans": [
      "Mark Zhao",
      "Dhruv Choudhary",
      "Devashish Tyagi",
      "Ajay Somani",
      "Max Kaplan",
      "Sung-Han Lin",
      "Sarunya Pumma",
      "Jongsoo Park",
      "Aarti Basant",
      "Niket Agarwal",
      "Carole-Jean Wu",
      "Christos Kozyrakis"
    ],
    "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-2211-05239v1",
        "title": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
        "url": "https://arxiv.org/abs/2211.05239v1"
      }
    ],
    "points": "70000",
    "published_grade": {
      "id": "zhao-et-al-recd-2022",
      "entry_id": "zhao-et-al-recd-2022",
      "band": "B",
      "modifier": "+",
      "target": {
        "record_id": "urn:uuid:eeb75c07-1597-4107-a466-2d1aca5997de",
        "version": 1,
        "content_hash": "sha256:54d01826b02f02fae945896ee878020712616ce11a00ab17370e99cad887590d"
      },
      "what_changed": "Coalesced session samples across logging, ETL and training, introducing inverse-keyed jagged tensors that share repeated feature values while retaining inverse mappings to their training examples.",
      "scope": "Coalesced session samples across logging, ETL and training, introducing inverse-keyed jagged tensors that share repeated feature values while retaining inverse mappings to their training examples. Deduplicates repeated features, not independent labels or arbitrary examples. Session grouping changes ordering and requires pipeline support; efficiency depends on actual session reuse and feature distributions. Reported storage, preprocessing and training gains are different metrics on selected industrial workloads.",
      "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:eeb75c07-1597-4107-a466-2d1aca5997de",
        "version": 1,
        "content_hash": "sha256:54d01826b02f02fae945896ee878020712616ce11a00ab17370e99cad887590d"
      },
      "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": "Mark Zhao",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
            "url": "https://arxiv.org/abs/2211.05239v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Dhruv Choudhary",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
            "url": "https://arxiv.org/abs/2211.05239v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Devashish Tyagi",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
            "url": "https://arxiv.org/abs/2211.05239v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Ajay Somani",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
            "url": "https://arxiv.org/abs/2211.05239v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Max Kaplan",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
            "url": "https://arxiv.org/abs/2211.05239v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Sung-Han Lin",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
            "url": "https://arxiv.org/abs/2211.05239v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Sarunya Pumma",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
            "url": "https://arxiv.org/abs/2211.05239v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Jongsoo Park",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
            "url": "https://arxiv.org/abs/2211.05239v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Aarti Basant",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
            "url": "https://arxiv.org/abs/2211.05239v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Niket Agarwal",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
            "url": "https://arxiv.org/abs/2211.05239v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Carole-Jean Wu",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
            "url": "https://arxiv.org/abs/2211.05239v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Christos Kozyrakis",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
            "url": "https://arxiv.org/abs/2211.05239v1"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
            "url": "https://arxiv.org/abs/2211.05239v1"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "Deduplicates repeated features, not independent labels or arbitrary examples. Session grouping changes ordering and requires pipeline support; efficiency depends on actual session reuse and feature distributions. Reported storage, preprocessing and training gains are different metrics on selected industrial workloads.",
        "domain": "Recommendation training infrastructure"
      },
      "sources": [
        {
          "citation": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
          "url": "https://arxiv.org/abs/2211.05239v1"
        }
      ],
      "statement": "# RecD: session-aware feature deduplication across recommendation training\n\nCoalesced session samples across logging, ETL and training, introducing inverse-keyed jagged tensors that share repeated feature values while retaining inverse mappings to their training examples.\n\n## Scientific scope\n\nDeduplicates repeated features, not independent labels or arbitrary examples. Session grouping changes ordering and requires pipeline support; efficiency depends on actual session reuse and feature distributions. Reported storage, preprocessing and training gains are different metrics on selected industrial workloads.\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\nMark Zhao; Dhruv Choudhary; Devashish Tyagi; Ajay Somani; Max Kaplan; Sung-Han Lin; Sarunya Pumma; Jongsoo Park; Aarti Basant; Niket Agarwal; Carole-Jean Wu; Christos Kozyrakis\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- [RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure](https://arxiv.org/abs/2211.05239v1)\n",
      "summary": "Coalesced session samples across logging, ETL and training, introducing inverse-keyed jagged tensors that share repeated feature values while retaining inverse mappings to their training examples.",
      "title": "RecD: session-aware feature deduplication across recommendation training"
    },
    "record_id": "urn:uuid:eeb75c07-1597-4107-a466-2d1aca5997de",
    "version": 1,
    "source_ref": {
      "record_id": "urn:uuid:eeb75c07-1597-4107-a466-2d1aca5997de",
      "version": 1,
      "content_hash": "sha256:54d01826b02f02fae945896ee878020712616ce11a00ab17370e99cad887590d"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:c3eb789404ed540bf756087e822b01f09240d711e6a9802ef53794093d29fd87"
  },
  "target": {
    "record_id": "urn:uuid:eeb75c07-1597-4107-a466-2d1aca5997de",
    "version": 1,
    "content_hash": "sha256:54d01826b02f02fae945896ee878020712616ce11a00ab17370e99cad887590d"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "zhao-et-al-recd-2022",
    "entry_id": "zhao-et-al-recd-2022",
    "revision": 1,
    "title": "RecD: session-aware feature deduplication across recommendation training",
    "context": "Recommendation training infrastructure",
    "track": "Human-designed research method or resource",
    "band": "B",
    "modifier": "+",
    "status": "PROVISIONAL_EDITORIAL",
    "assessed_on": "2026-10-16",
    "evidence_cutoff": "2026-10-16",
    "achievement_state": "SUPPORTED_SCOPED_RESULT",
    "what_changed": "Coalesced session samples across logging, ETL and training, introducing inverse-keyed jagged tensors that share repeated feature values while retaining inverse mappings to their training examples.",
    "scope": "Coalesced session samples across logging, ETL and training, introducing inverse-keyed jagged tensors that share repeated feature values while retaining inverse mappings to their training examples. Deduplicates repeated features, not independent labels or arbitrary examples. Session grouping changes ordering and requires pipeline support; efficiency depends on actual session reuse and feature distributions. Reported storage, preprocessing and training gains are different metrics on selected industrial workloads.",
    "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. Deduplicates repeated features, not independent labels or arbitrary examples. Session grouping changes ordering and requires pipeline support; efficiency depends on actual session reuse and feature distributions. Reported storage, preprocessing and training gains are different metrics on selected industrial workloads.",
    "references": [
      {
        "title": "RecD: Deduplication for End-to-End Deep Learning Recommendation Model Training Infrastructure",
        "url": "https://arxiv.org/abs/2211.05239v1",
        "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:eeb75c07-1597-4107-a466-2d1aca5997de",
      "version": 1,
      "content_hash": "sha256:54d01826b02f02fae945896ee878020712616ce11a00ab17370e99cad887590d"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "70000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": null,
  "reading_status": "SOURCE_LINKED_NOTE"
}