{
  "manifest": {
    "schema_version": "1.0",
    "design": {
      "schema_version": "1.0",
      "frozen_date": "2026-09-12",
      "scope": "Exploratory organization and retrieval within the existing identity-hit OCR passage pool. Not a representative article corpus, local-reference classification, or test of recovery outside the identity filter.",
      "source_sha256": "0f093fc95ed337794384cfaeef51941a7c5058266bfa3949a8e2343bf7b4f688",
      "model": "sentence-transformers/all-MiniLM-L6-v2",
      "revision": "1110a243fdf4706b3f48f1d95db1a4f5529b4d41",
      "max_tokens": 256,
      "pooling": "attention-mask mean pooling, L2 normalized",
      "seed": 42,
      "clusters": 16,
      "layout": {
        "method": "UMAP",
        "n_neighbors": 30,
        "min_dist": 0.25,
        "metric": "cosine",
        "random_state": 42
      },
      "stability": "Compare 16-cluster MiniBatchKMeans assignments at seeds 42 and 17. Inspect centroid-nearest examples and lexical descriptors. This is a sensitivity check, not a relevance evaluation.",
      "queries": [
        {
          "id": "school",
          "label": "Going to school",
          "query": "Chinese children attending schools and facing exclusion from education in Mississippi."
        },
        {
          "id": "business",
          "label": "Earning a living",
          "query": "Chinese residents running local grocery stores, laundries, shops and businesses in Mississippi towns."
        },
        {
          "id": "farmer",
          "label": "A farmer near Port Gibson",
          "query": "Quong Thomas, a Chinese farmer near Port Gibson in Mississippi, grows cotton and has savings in a bank."
        },
        {
          "id": "rights",
          "label": "Belonging and exclusion",
          "query": "Chinese immigrants, citizenship, exclusion laws, racial discrimination and the right to live in the United States."
        },
        {
          "id": "world",
          "label": "News from elsewhere",
          "query": "War, diplomacy, rebellion and foreign troops in China."
        }
      ],
      "retrieval": "Cosine ranking in original 384-dimensional embeddings; compare word TF-IDF unigram/bigram cosine with min_df=2, max_features=60000, sublinear_tf=true and sklearn English stop words. Return at most one passage per saved page key for the first 100 pages per method/query. Do not call ranking differences improved relevance.",
      "publication": "Expose candidate OCR as candidate OCR, separate from accepted source transcriptions. No model accuracy, trend, prevalence, validated story-family or biographical claim is licensed. Guided reviewed examples preserve their own existing review status."
    },
    "design_sha256": "522b1301906596fb3e936a43c33e3dde5e5e3f9babd08181d5abdb4e24fedd47",
    "script_sha256": "7e83346c363d06e4cc5235113b391ce3cd742e36b5592641912552aa69e8feaf",
    "input_sha256": "0f093fc95ed337794384cfaeef51941a7c5058266bfa3949a8e2343bf7b4f688",
    "passages": 46449,
    "page_keys": 24346,
    "distinct_texts": 46012,
    "titles": 220,
    "year_range": [
      1820,
      1963
    ],
    "review_status": "Unreviewed OCR candidates. Guided sources retain their separate AI image review.",
    "cluster_seed_adjusted_rand": 0.3842222609802737,
    "embedding_sha256": "6037318b6a859981f376b0400c63dcceb58fb7dcd36dda3354daaaf97b6f3407",
    "model_file_sha256": {
      "model.safetensors": "53aa51172d142c89d9012cce15ae4d6cc0ca6895895114379cacb4fab128d9db",
      "tokenizer_config.json": "acb92769e8195aabd29b7b2137a9e6d6e25c476a4f15aa4355c233426c61576b",
      "special_tokens_map.json": "303df45a03609e4ead04bc3dc1536d0ab19b5358db685b6f3da123d05ec200e3",
      "config.json": "953f9c0d463486b10a6871cc2fd59f223b2c70184f49815e7efbcab5d8908b41",
      "tokenizer.json": "be50c3628f2bf5bb5e3a7f17b1f74611b2561a3a27eeab05e5aa30f411572037",
      "README.md": "dcd602d2fd35c203a247304a06fec6654a12f7941b739f9221a064fe8dc3b7f0",
      "sentence_bert_config.json": "fc1993fde0a95c24ec6c022539d41cf6e2f7c9721e5415d6fb6897472a9cd4b7",
      "vocab.txt": "07eced375cec144d27c900241f3e339478dec958f92fddbc551f295c992038a3",
      "modules.json": "84e40c8e006c9b1d6c122e02cba9b02458120b5fb0c87b746c41e0207cf642cf",
      "1_Pooling/config.json": "4be450dde3b0273bb9787637cfbd28fe04a7ba6ab9d36ac48e92b11e350ffc23"
    },
    "runtime": {
      "python": "3.12.13",
      "torch": "2.9.1",
      "numpy": "2.4.0",
      "sklearn": "1.8.0",
      "umap": "0.5.12",
      "device": "mps"
    },
    "seconds": 98.58
  },
  "questions": [
    {
      "id": "school",
      "label": "Going to school",
      "query": "Chinese children attending schools and facing exclusion from education in Mississippi.",
      "top20_page_overlap": 4
    },
    {
      "id": "business",
      "label": "Earning a living",
      "query": "Chinese residents running local grocery stores, laundries, shops and businesses in Mississippi towns.",
      "top20_page_overlap": 2
    },
    {
      "id": "farmer",
      "label": "A farmer near Port Gibson",
      "query": "Quong Thomas, a Chinese farmer near Port Gibson in Mississippi, grows cotton and has savings in a bank.",
      "top20_page_overlap": 11
    },
    {
      "id": "rights",
      "label": "Belonging and exclusion",
      "query": "Chinese immigrants, citizenship, exclusion laws, racial discrimination and the right to live in the United States.",
      "top20_page_overlap": 0
    },
    {
      "id": "world",
      "label": "News from elsewhere",
      "query": "War, diplomacy, rebellion and foreign troops in China.",
      "top20_page_overlap": 0
    }
  ],
  "editorial_decision": "Do not expose seed-sensitive clustering as historical categories. Use fixed queries and source reading.",
  "source": "Exploratory organization and retrieval within the existing identity-hit OCR passage pool. Not a representative article corpus, local-reference classification, or test of recovery outside the identity filter."
}