{
  "slug": "resemblance-detection",
  "term": "Resemblance detection",
  "aliases": [
    "Chunk resemblance detection",
    "Similarity-based base selection"
  ],
  "definition": "Resemblance detection identifies similar, potentially nonidentical data objects as candidates for encoding one relative to another.",
  "note": "Selection: A compact similarity summary narrows the search for a useful base without requiring the target to be a new version of the same named file. REBL first removes duplicate chunks, then searches for resembling chunks and applies delta encoding. [rebl]\n\nSuper-fingerprints: Grouping several features into a hash-table key reduces candidate-matching work. More restrictive matching can miss useful bases; a selected candidate can still compress poorly. Similarity selection is separate from the correctness of reconstructing the target bytes. [rebl]\n\nRecovery: A stored delta creates a dependency on its selected base. REBL assigns a chunk either as a reference or as an encoded version, rather than repeatedly using encoded versions as new bases. [rebl]",
  "related": [
    "finesse",
    "delta-compression",
    "deduplication",
    "content-defined-chunking",
    "minhash"
  ],
  "citations": [
    {
      "id": "rebl",
      "title": "Kulkarni, Douglis, LaVoie and Tracey. Redundancy Elimination Within Large Collections of Files. USENIX ATC 2004, sections 2-4.",
      "url": "https://atc.usenix.org/legacy/events/usenix04/tech/general/full_papers/kulkarni/kulkarni_html/index.html",
      "accessed": "2026-11-22"
    }
  ],
  "status": "research",
  "advantages": [],
  "limitations": [],
  "supersedes": [],
  "supersededBy": [],
  "firstSeen": null,
  "lastSeen": null,
  "updated": "2026-11-22",
  "provenance": [
    "Source synthesis, not a similarity detector implementation or a measured compression result. [rebl]"
  ]
}
