roseetal2019nothingstakeknowledge
/data/papers/roseetal2019nothingstakeknowledge/out/bundle.json
{
  "created_at": "2026-01-19T10:55:33.294660+00:00",
  "paper_dir": "papers/roseetal2019nothingstakeknowledge",
  "pdf": {
    "path": "papers/roseetal2019nothingstakeknowledge/pdf/paper.pdf"
  },
  "tool_versions": {
    "python": "3.14.2",
    "meta-extract": "0.1.0",
    "requests": "2.32.5",
    "lxml": "6.0.2",
    "pandas": "2.3.3",
    "pypdf": "5.9.0",
    "pypdfium2": "5.3.0",
    "pillow": "12.1.0",
    "grobid-client-python": "0.1.4",
    "camelot-py": "1.0.9",
    "tabula-py": "2.10.0"
  },
  "status": "ok",
  "grobid": {
    "url": "http://localhost:8070",
    "tei_path": "papers/roseetal2019nothingstakeknowledge/out/tei.xml"
  },
  "text": {
    "path": "papers/roseetal2019nothingstakeknowledge/out/text.txt"
  },
  "tables": {
    "out_dir": "papers/roseetal2019nothingstakeknowledge/out/tables",
    "camelot": {
      "stream": [
        {
          "page": "9",
          "index": 1,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/camelot_stream_p9_t1.csv",
          "shape": [
            28,
            6
          ],
          "stats": {
            "rows": 28,
            "cols": 6,
            "non_empty_cells": 138,
            "non_empty_ratio": 0.8214285714285714,
            "filled_rows": 22,
            "filled_cols": 6,
            "filled_rows_ratio": 0.7857142857142857,
            "numeric_like_ratio": 0.15217391304347827
          },
          "parsing_report": {
            "accuracy": 99.9,
            "whitespace": 17.86,
            "order": 1,
            "page": 9
          }
        },
        {
          "page": "10",
          "index": 2,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/camelot_stream_p10_t2.csv",
          "shape": [
            27,
            5
          ],
          "stats": {
            "rows": 27,
            "cols": 5,
            "non_empty_cells": 106,
            "non_empty_ratio": 0.7851851851851852,
            "filled_rows": 20,
            "filled_cols": 5,
            "filled_rows_ratio": 0.7407407407407407,
            "numeric_like_ratio": 0.25471698113207547
          },
          "parsing_report": {
            "accuracy": 99.93,
            "whitespace": 21.48,
            "order": 1,
            "page": 10
          }
        },
        {
          "page": "13",
          "index": 3,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/camelot_stream_p13_t3.csv",
          "shape": [
            26,
            5
          ],
          "stats": {
            "rows": 26,
            "cols": 5,
            "non_empty_cells": 106,
            "non_empty_ratio": 0.8153846153846154,
            "filled_rows": 20,
            "filled_cols": 5,
            "filled_rows_ratio": 0.7692307692307693,
            "numeric_like_ratio": 0.12264150943396226
          },
          "parsing_report": {
            "accuracy": 99.9,
            "whitespace": 18.46,
            "order": 1,
            "page": 13
          }
        },
        {
          "page": "14",
          "index": 4,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/camelot_stream_p14_t4.csv",
          "shape": [
            27,
            5
          ],
          "stats": {
            "rows": 27,
            "cols": 5,
            "non_empty_cells": 106,
            "non_empty_ratio": 0.7851851851851852,
            "filled_rows": 20,
            "filled_cols": 5,
            "filled_rows_ratio": 0.7407407407407407,
            "numeric_like_ratio": 0.27358490566037735
          },
          "parsing_report": {
            "accuracy": 99.93,
            "whitespace": 21.48,
            "order": 1,
            "page": 14
          }
        },
        {
          "page": "17",
          "index": 5,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/camelot_stream_p17_t5.csv",
          "shape": [
            26,
            5
          ],
          "stats": {
            "rows": 26,
            "cols": 5,
            "non_empty_cells": 106,
            "non_empty_ratio": 0.8153846153846154,
            "filled_rows": 20,
            "filled_cols": 5,
            "filled_rows_ratio": 0.7692307692307693,
            "numeric_like_ratio": 0.1320754716981132
          },
          "parsing_report": {
            "accuracy": 99.9,
            "whitespace": 18.46,
            "order": 1,
            "page": 17
          }
        },
        {
          "page": "18",
          "index": 6,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/camelot_stream_p18_t6.csv",
          "shape": [
            27,
            5
          ],
          "stats": {
            "rows": 27,
            "cols": 5,
            "non_empty_cells": 106,
            "non_empty_ratio": 0.7851851851851852,
            "filled_rows": 20,
            "filled_cols": 5,
            "filled_rows_ratio": 0.7407407407407407,
            "numeric_like_ratio": 0.25471698113207547
          },
          "parsing_report": {
            "accuracy": 99.93,
            "whitespace": 21.48,
            "order": 1,
            "page": 18
          }
        },
        {
          "page": "21",
          "index": 7,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/camelot_stream_p21_t7.csv",
          "shape": [
            26,
            5
          ],
          "stats": {
            "rows": 26,
            "cols": 5,
            "non_empty_cells": 106,
            "non_empty_ratio": 0.8153846153846154,
            "filled_rows": 20,
            "filled_cols": 5,
            "filled_rows_ratio": 0.7692307692307693,
            "numeric_like_ratio": 0.1792452830188679
          },
          "parsing_report": {
            "accuracy": 99.9,
            "whitespace": 18.46,
            "order": 1,
            "page": 21
          }
        }
      ],
      "lattice": [],
      "filtered_out": {
        "stream": 17,
        "lattice": 0
      },
      "filtered_reasons": {
        "stream": {
          "min_cols": 13,
          "filled_rows": 2,
          "filled_rows_ratio": 1,
          "sparse": 1
        },
        "lattice": {}
      },
      "filter": {
        "min_rows": 2,
        "min_cols": 2,
        "min_non_empty_ratio": 0.2,
        "min_filled_rows": 2,
        "min_filled_cols": 2,
        "min_filled_rows_ratio": 0.5,
        "two_col_min_numeric_like_ratio": 0.35
      },
      "errors": []
    },
    "tabula": {
      "lattice": [],
      "stream": [
        {
          "page": 9,
          "index": 1,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/tabula_stream_p9_t1.csv",
          "shape": [
            27,
            6
          ],
          "stats": {
            "rows": 27,
            "cols": 6,
            "non_empty_cells": 132,
            "non_empty_ratio": 0.8148148148148148,
            "filled_rows": 21,
            "filled_cols": 6,
            "filled_rows_ratio": 0.7777777777777778,
            "numeric_like_ratio": 0.1590909090909091
          }
        },
        {
          "page": 10,
          "index": 2,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/tabula_stream_p10_t2.csv",
          "shape": [
            25,
            5
          ],
          "stats": {
            "rows": 25,
            "cols": 5,
            "non_empty_cells": 101,
            "non_empty_ratio": 0.808,
            "filled_rows": 19,
            "filled_cols": 5,
            "filled_rows_ratio": 0.76,
            "numeric_like_ratio": 0.5643564356435643
          }
        },
        {
          "page": 12,
          "index": 3,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/tabula_stream_p12_t3.csv",
          "shape": [
            10,
            7
          ],
          "stats": {
            "rows": 10,
            "cols": 7,
            "non_empty_cells": 43,
            "non_empty_ratio": 0.6142857142857143,
            "filled_rows": 8,
            "filled_cols": 7,
            "filled_rows_ratio": 0.8,
            "numeric_like_ratio": 0.0
          },
          "area": [
            466.56,
            25.919999999999998,
            635.04,
            406.08
          ]
        },
        {
          "page": 13,
          "index": 4,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/tabula_stream_p13_t4.csv",
          "shape": [
            25,
            5
          ],
          "stats": {
            "rows": 25,
            "cols": 5,
            "non_empty_cells": 101,
            "non_empty_ratio": 0.808,
            "filled_rows": 19,
            "filled_cols": 5,
            "filled_rows_ratio": 0.76,
            "numeric_like_ratio": 0.12871287128712872
          }
        },
        {
          "page": 14,
          "index": 5,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/tabula_stream_p14_t5.csv",
          "shape": [
            25,
            5
          ],
          "stats": {
            "rows": 25,
            "cols": 5,
            "non_empty_cells": 101,
            "non_empty_ratio": 0.808,
            "filled_rows": 19,
            "filled_cols": 5,
            "filled_rows_ratio": 0.76,
            "numeric_like_ratio": 0.5643564356435643
          }
        },
        {
          "page": 17,
          "index": 6,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/tabula_stream_p17_t6.csv",
          "shape": [
            25,
            5
          ],
          "stats": {
            "rows": 25,
            "cols": 5,
            "non_empty_cells": 101,
            "non_empty_ratio": 0.808,
            "filled_rows": 19,
            "filled_cols": 5,
            "filled_rows_ratio": 0.76,
            "numeric_like_ratio": 0.13861386138613863
          }
        },
        {
          "page": 18,
          "index": 7,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/tabula_stream_p18_t7.csv",
          "shape": [
            25,
            5
          ],
          "stats": {
            "rows": 25,
            "cols": 5,
            "non_empty_cells": 101,
            "non_empty_ratio": 0.808,
            "filled_rows": 19,
            "filled_cols": 5,
            "filled_rows_ratio": 0.76,
            "numeric_like_ratio": 0.7524752475247525
          }
        },
        {
          "page": 21,
          "index": 8,
          "csv_path": "papers/roseetal2019nothingstakeknowledge/out/tables/tabula_stream_p21_t8.csv",
          "shape": [
            25,
            5
          ],
          "stats": {
            "rows": 25,
            "cols": 5,
            "non_empty_cells": 101,
            "non_empty_ratio": 0.808,
            "filled_rows": 19,
            "filled_cols": 5,
            "filled_rows_ratio": 0.76,
            "numeric_like_ratio": 0.18811881188118812
          }
        }
      ],
      "filtered_out": {
        "lattice": 1,
        "stream": 2
      },
      "filtered_reasons": {
        "lattice": {
          "min_rows": 1
        },
        "stream": {
          "min_cols": 1,
          "min_rows": 1
        }
      },
      "filter": {
        "min_rows": 2,
        "min_cols": 2,
        "min_non_empty_ratio": 0.2,
        "min_filled_rows": 2,
        "min_filled_cols": 2,
        "min_filled_rows_ratio": 0.5,
        "two_col_min_numeric_like_ratio": 0.35
      },
      "errors": [],
      "ignored_errors": {},
      "pages": "all"
    }
  },
  "images": {
    "dir": "papers/roseetal2019nothingstakeknowledge/out/images",
    "manifest": "papers/roseetal2019nothingstakeknowledge/out/images/manifest.json",
    "embedded": {
      "count": 10
    },
    "rendered_pages": {
      "count": 7,
      "pages": [
        1,
        11,
        12,
        15,
        16,
        19,
        20
      ]
    },
    "figures": {
      "count": 15,
      "pages": [
        11,
        12,
        15,
        16,
        19,
        20
      ],
      "rendered_pages": [
        11,
        12,
        15,
        16,
        19,
        20
      ]
    }
  },
  "errors": [],
  "markdown": {
    "path": "papers/roseetal2019nothingstakeknowledge/out/fulltext.md"
  }
}