Aggregate decisions.
Never average unlike distances.

{
  "module": "Compo 29.04",
  "canonical_main_title": "Multi-channel groups and tests",
  "topic": "29_СХОДСТВО_И_ПОВТОРЯЕМОСТЬ",
  "scope": "EXPLICIT_SLOT_WITHIN_CHANNEL_AND_CHANNEL_ACROSS_PAIR_AGGREGATION_OVER_IMMUTABLE_29_01_29_02_29_03_RELATIONS_WITH_SIMILAR_LINK_COMPONENT_CLUSTERING_ONLY",
  "main_topic_contract": {
    "topic_title": "Сходство и повторяемость",
    "topic_goal": "Измеряет сходство по независимым каналам — shape, size, color, tone, orientation, texture/style — и формирует similarity groups без потери источника сигнала.",
    "expected_output": "SimilarityRelation[], feature distances, similarity clusters",
    "planned_files": [
      "29.00 Similarity feature schema",
      "29.01 Shape/size similarity",
      "29.02 Color/tone similarity",
      "29.03 Orientation/texture/style similarity",
      "29.04 Multi-channel groups and tests"
    ]
  },
  "ownership": {
    "owner": "Compo29.04",
    "owns": [
      "SimilaritySlotMatchRule",
      "SimilarityChannelAggregationRule",
      "SimilarityMultiChannelPolicy",
      "SimilaritySlotMatchDescriptor",
      "SimilarityChannelSupportDescriptor",
      "SimilarityPairSupportDescriptor",
      "SimilarityPairDecision",
      "SimilarityClusterCandidate",
      "SimilarityClusterSet",
      "SimilarityMultiChannelAttachment",
      "slot-threshold classification",
      "slot-weight aggregation inside channel",
      "channel-weight aggregation across channels",
      "coverage gating",
      "SIMILAR_LINK decision",
      "NON_SIMILAR_LINK decision",
      "N_A insufficient-coverage preservation",
      "connected components over SIMILAR_LINK",
      "singleton clusters",
      "cluster evidence lineage",
      "bounded explicit relation-union processing",
      "non-mutating correction request"
    ],
    "does_not_own": [
      "29.00 feature schema",
      "29.01 SHAPE/SIZE distances",
      "29.02 COLOR/TONE distances",
      "29.03 ORIENTATION/TEXTURE_STYLE distances",
      "raw distance recomputation",
      "automatic raw-distance normalization",
      "automatic threshold selection",
      "automatic channel weighting",
      "automatic slot weighting",
      "full Cartesian pair generation",
      "raw cross-channel distance averaging",
      "clique completion",
      "transitive similarity truth",
      "measured human similarity judgment",
      "Topic28 proximity grouping",
      "Topic27 attention flow",
      "repair generation",
      "universal similarity quality",
      "universal beautyScore"
    ]
  },
  "pair_decisions": [
    "SIMILAR_LINK",
    "NON_SIMILAR_LINK"
  ],
  "aggregation_semantics": {
    "slot": "distance <= explicit maxDistance => MATCH; missing relation => N_A",
    "channel": "slotCoverageShare + slotMatchShare; one outer channelWeight regardless of slot count",
    "pair": "channelCoverageShare + weightedChannelMatchShare",
    "pair_N_A": "coverage below explicit minimum",
    "cluster": "connected components of SIMILAR_LINK only",
    "clique_claim": false,
    "transitivity_claim": false,
    "perceptual_truth_claim": false,
    "raw_distance_averaging": false
  },
  "policy": {
    "max_relations": [
      0,
      393216
    ],
    "max_pair_descriptors": [
      0,
      131072
    ],
    "max_clusters": [
      1,
      4096
    ],
    "overflow": "N_A_NO_TRUNCATION",
    "auto_threshold_selection": false,
    "auto_slot_weight_selection": false,
    "auto_channel_weight_selection": false,
    "auto_missing_relation_imputation": false,
    "auto_pair_cartesian": false,
    "auto_raw_distance_normalization": false,
    "auto_raw_distance_averaging": false,
    "auto_clique_completion": false,
    "auto_transitive_similarity_claim": false,
    "auto_proximity_fusion": false,
    "auto_attention_fusion": false,
    "auto_scoring": false
  },
  "status_semantics": {
    "OK": "valid rules and upstream relations resolve deterministic pair-support descriptors and SIMILAR_LINK components",
    "N_A": "pair lacks configured channel coverage, required relation lineage is absent, or bounds overflow",
    "INVALID": "malformed rule/policy/relation, duplicate identity, negative distance, invalid threshold/weight, or malformed upstream set",
    "zero": "distance 0, match share 0 and coverage 0 can be valid; N_A remains distinct"
  },
  "firewalls": [
    "raw distance != cross-channel score",
    "channel threshold != universal similarity threshold",
    "slot match != channel match",
    "channel match != pair similarity fact",
    "weighted channel match share != empirical probability",
    "SIMILAR_LINK != measured human similarity",
    "NON_SIMILAR_LINK != semantic unrelatedness",
    "N_A != NON_SIMILAR_LINK",
    "connected component != clique",
    "cluster membership != every member pair being similar",
    "connected component != transitive similarity law",
    "similarity cluster != perceptual truth",
    "slot proliferation cannot increase outer channel weight",
    "multiple texture/style slots do not create extra channels",
    "missing relation != mismatch",
    "distance zero != N_A",
    "pair union != full Cartesian product",
    "Topic29 similarity cluster != Topic28 proximity group",
    "Topic29 SIMILAR_LINK != Topic28 neighbor relation",
    "Topic29 similarity cluster != Topic27 attention path",
    "N_A != 0",
    "INVALID != 0",
    "descriptor != score",
    "local != global",
    "geometry != interpretation",
    "heuristic != law",
    "no universal similarity threshold",
    "no universal beautyScore"
  ],
  "owns_metrics": [],
  "independent_numeric_votes": 0,
  "topic29ProgressAfterThisModule": "5/5 local",
  "next_step": "COMPO_TOPIC_29_CLOSING_AUDIT_FINAL — separate artifact",
  "topic30_started": false,
  "publication": "NOT_VERIFIED"
}