Switch to one flat set of shards

README.md

@@ -5,15 +5,7 @@ configs:
   - config_name: default
     data_files:
       - split: train
-        path: data/*/*.jsonl.gz
-  - config_name: asserted
-    data_files:
-      - split: train
-        path: data/asserted/*.jsonl.gz
-  - config_name: mined
-    data_files:
-      - split: train
-        path: data/mined/*.jsonl.gz
+        path: data/*.jsonl.gz
 tags:
   - datacite
   - crossref
@@ -37,7 +29,7 @@ Each record targets a cited DataCite DOI in `doi`. It inserts an `IsCitedBy` rel
 
 ## Contents
 
-The dataset contains **7,339,071 citations of 1,705,826 DataCite DOIs**, stored in 31 gzip-compressed JSONL shards in two partitions.
+The dataset contains **7,339,071 citations of 1,705,826 DataCite DOIs**, stored in 30 gzip-compressed JSONL shards.
 
 | Measure | Count |
 | --- | ---: |
@@ -46,23 +38,14 @@ The dataset contains **7,339,071 citations of 1,705,826 DataCite DOIs**, stored
 | Source citing-work entries | 7,339,071 |
 | Source reference matches | 7,385,186 |
 | Output records | 7,339,071 |
-| Records in `asserted` | 1,269,805 |
-| `asserted` records with strongest provenance `publisher` | 1,192,204 |
-| `asserted` records with strongest provenance `crossref` | 77,601 |
-| Records in `mined` | 6,069,266 |
+| Records whose citing work's strongest provenance is `publisher` | 1,192,204 |
+| Records whose citing work's strongest provenance is `crossref` | 77,601 |
+| Records whose citing work's strongest provenance is `mined` | 6,069,266 |
 | Records whose cited DOI is an arXiv DOI | 5,999,691 |
 
 Each source row lists every citing work once, so each pair of cited DOI and citing DOI yields one record. A citing work can match the cited DOI in more than one of its references. That is why source reference matches exceed records.
 
-The enrichment schema has no field for provenance, so the file layout keeps it. Each source citing-work entry carries the strongest provenance among its matches, ranked `publisher`, then `crossref`, then `mined`. The conversion writes `publisher` and `crossref` entries to `data/asserted/` and `mined` entries to `data/mined/`. Any other value stops the conversion.
-
-| Config | Files | Records |
-| --- | --- | ---: |
-| `default` | `data/*/*.jsonl.gz` | 7,339,071 |
-| `asserted` | `data/asserted/*.jsonl.gz` | 1,269,805 |
-| `mined` | `data/mined/*.jsonl.gz` | 6,069,266 |
-
-Each pair appears in exactly one partition. `asserted` holds the same 1,269,805 pairs as the source `asserted` configuration. `mined` holds the pairs whose strongest provenance is `mined`. The source `mined` configuration lists 6,070,488 pairs. 1,222 of those pairs have a mined match alongside a `publisher` or `crossref` match, so this dataset places them only in `asserted`.
+The source gives each citing-work entry the strongest provenance among its matches, ranked `publisher`, then `crossref`, then `mined`. The provenance rows above count that value for information. The records do not carry it.
 
 ## Mapping
 
@@ -71,7 +54,7 @@ Each pair appears in exactly one partition. `asserted` holds the same 1,269,805
 - Each `cited_by[].doi` maps to `enrichedValue.relatedIdentifier`. The value is a bare DOI name, as in the DataCite 4.7 relatedIdentifier example, not a `https://doi.org/` URL.
 - `relatedIdentifierType` is always `DOI`. `relationType` is always `IsCitedBy`.
 - `resourceTypeGeneral` is omitted because the source does not record the citing work's type.
-- The source match details (`match_type`, `raw_match`, and `reference`) are not carried. Only the partition keeps the provenance.
+- The records do not carry the source match details (`match_type`, `raw_match`, and `reference`) or the citing work's `provenance`.
 - Both DOIs must pass the comet-enrich `is_valid_doi_name` check and be lowercase. A violation stops the conversion instead of dropping the record. The pinned source has none.
 - A repeated pair of cited DOI and citing DOI stops the conversion. The pinned source has none.
 
@@ -85,12 +68,12 @@ DOIs are copied as the source supplies them, already lowercase. Four cited DOIs
   "action": "insert",
   "field": "relatedIdentifiers",
   "enrichedValue": {
-    "relatedIdentifier": "10.1002/9781394372539.ch6",
+    "relatedIdentifier": "10.1001/jamadermatol.2022.5415",
     "relatedIdentifierType": "DOI",
     "relationType": "IsCitedBy"
   },
   "sourceId": "",
-  "contentKey": "8f221c7f80f2b7bad183b55359e24bb7"
+  "contentKey": "63f52c05c46e67981b4f1f5157606c91"
 }
 ```
 
@@ -107,11 +90,11 @@ This follows the current [COMET content-key implementation](https://github.com/c
 
 ## Use
 
-Download the files under `data/` for direct enrichment ingestion. Each decompressed line is one schema-valid insertion record. To ingest only citations whose DOI the publisher or Crossref asserted, use `data/asserted/` or the `asserted` configuration.
+Download the files under `data/` for direct enrichment ingestion. Each decompressed line is one schema-valid insertion record. The records do not carry provenance. Readers who want only citations whose DOI the publisher or Crossref asserted can use the `asserted` configuration of the source dataset.
 
 These are proposed insertions. This dataset does not apply them to DataCite records. It does not compare them with the relatedIdentifiers already deposited in DataCite, so some records may duplicate an existing relation.
 
-Each configuration has one `train` split. Every record has all three `enrichedValue` properties, so loading through Arrow adds no null properties.
+The `default` configuration has one `train` split. Every record has all three `enrichedValue` properties, so loading through Arrow adds no null properties.
 
 ## Reproduce and validate
 
@@ -134,10 +117,10 @@ python verify_source.py rebuilt --source-dir source/data/all
 
 To fill `sourceId` once the project has a DOI, add `--source-id <doi>` to the `convert.py` command. The data files change, but every `contentKey` stays the same.
 
-`convert.py` checks the SHA-256 of the pinned pipeline manifest, then the size and SHA-256 of each `all` parquet file that the manifest lists, before it reads any rows. It validates every emitted record against the pinned enrichment schema and a related-identifier subset schema. It also checks its source totals against the counts in the pipeline manifest. `validate.py` reads all output shards and checks their SHA-256 checksums, both schemas, and both DOIs. It independently recomputes content keys and checks global key uniqueness and per-partition totals. `verify_source.py` compares every output tuple of cited DOI, citing DOI, and partition with a separate SQL projection of the source. That projection derives the partition from the per-match provenances, not from the per-work `provenance` field that `convert.py` reads. Results are in `validation_report.json`. Two conversions of the pinned source produced byte-identical data files.
+`convert.py` checks the SHA-256 of the pinned pipeline manifest, then the size and SHA-256 of each `all` parquet file that the manifest lists, before it reads any rows. It validates every emitted record against the pinned enrichment schema and a related-identifier subset schema. It also checks its source totals against the counts in the pipeline manifest. `validate.py` reads all output shards and checks their SHA-256 checksums, both schemas, and both DOIs. It independently recomputes content keys and checks global key uniqueness and the manifest totals. `verify_source.py` compares every output pair of cited DOI and citing DOI with a separate SQL projection of the source, in both directions, and checks that the counts match. Results are in `validation_report.json`. Two conversions of the pinned source produced byte-identical data files.
 
-The source revision is `65642a869543640f2965bcf9a7652f49fe88ef93`, built by extractor version `2.2.0` from the Crossref snapshot and DataCite release `2026-07`. The COMET schema revision is `abd6a9acd3fc70d25f703ea5be3c0fc62e0b6a61`. `provenance/` holds the schema snapshot, the related-identifier subset schema, the source pipeline manifest, the independent SQL projection, and four example records from each partition.
+The source revision is `65642a869543640f2965bcf9a7652f49fe88ef93`, built by extractor version `2.2.0` from the Crossref snapshot and DataCite release `2026-07`. The COMET schema revision is `abd6a9acd3fc70d25f703ea5be3c0fc62e0b6a61`. `provenance/` holds the schema snapshot, the related-identifier subset schema, the source pipeline manifest, the independent SQL projection, and the first four records.
 
 ## Limitations
 
-Citations reflect one Crossref snapshot and one DataCite release and keep the source's matching errors. The source notes that `mined` matches come from noisy reference text and may be lower quality. In 5,999,691 records (81.75%), the cited DOI is an arXiv DOI with the prefix `10.48550/`. This conversion does not check the relatedIdentifiers already deposited in DataCite, and it does not record the citing work's resource type. The source is released under CC0 1.0, and so is this conversion.
+Citations reflect one Crossref snapshot and one DataCite release and keep the source's matching errors. The source notes that `mined` matches come from noisy reference text and may be lower quality. In 6,069,266 records, the citing work's strongest provenance is `mined`. In 5,999,691 records (81.75%), the cited DOI is an arXiv DOI with the prefix `10.48550/`. This conversion does not check the relatedIdentifiers already deposited in DataCite, and it does not record the citing work's resource type. The source is released under CC0 1.0, and so is this conversion.

convert.py

@@ -27,9 +27,6 @@ SCHEMA_REVISION = "abd6a9acd3fc70d25f703ea5be3c0fc62e0b6a61"
 SCHEMA_SHA256 = "8a46a45770572b68261aa4045641006e22d56cc652ae3ec9242718a091ff7c28"
 METHOD = "crossref-datacite-citations"
 SCHEMA_PATH = Path(__file__).parent / "provenance/enrichment_input_schema.json"
-# The enrichment schema cannot carry provenance, so the file layout does.
-# Keys are the source's per-citing-work strongest provenance.
-PARTITIONS = {"publisher": "asserted", "crossref": "asserted", "mined": "mined"}
 DOI_NAME = re.compile(r"\A10\.[0-9]+(?:\.[0-9]+)*/.+\Z")
 # Rust's char::is_control is general category Cc, which is exactly these code points.
 CONTROL = re.compile(r"[\x00-\x1f\x7f-\x9f]")
@@ -79,11 +76,10 @@ def gzip_text(path: Path):
 
 
 class ShardWriter:
-    """Numbered gzip JSONL shards for one partition."""
+    """Numbered gzip JSONL shards of rows_per_shard records under data/."""
 
-    def __init__(self, output: Path, partition: str, rows_per_shard: int):
+    def __init__(self, output: Path, rows_per_shard: int):
         self.output = output
-        self.directory = output / "data" / partition
         self.rows_per_shard = rows_per_shard
         self.files = []
         self.stream = None
@@ -93,8 +89,7 @@ class ShardWriter:
     def write(self, line: str):
         if self.stream is None or self.rows == self.rows_per_shard:
             self.close()
-            self.directory.mkdir(parents=True, exist_ok=True)
-            self.path = self.directory / f"part_{len(self.files):04d}.jsonl.gz"
+            self.path = self.output / "data" / f"part_{len(self.files):04d}.jsonl.gz"
             self.stream = gzip_text(self.path)
             self.rows = 0
         self.stream.write(line)
@@ -145,9 +140,10 @@ def main():
     pipeline, inputs = pinned_source(args.source_dir)
     output = args.output_dir
     data = output / "data"
-    if data.exists() and any(data.iterdir()):
+    data.mkdir(parents=True, exist_ok=True)
+    if any(data.iterdir()):
         raise ValueError("Output data directory is not empty")
-    (output / "provenance").mkdir(parents=True, exist_ok=True)
+    (output / "provenance").mkdir(exist_ok=True)
     validate_enrichment = fastjsonschema.compile(json.loads(SCHEMA_PATH.read_text()))
     validate_value = fastjsonschema.compile(VALUE_SCHEMA)
     (output / "provenance/related_identifier_subset_schema.json").write_text(
@@ -155,14 +151,12 @@ def main():
     shutil.copyfile(args.source_dir / PIPELINE_MANIFEST,
                     output / "provenance/source_pipeline_manifest.json")
     counts = Counter()
-    by_partition = Counter()
     by_provenance = Counter()
     seen_keys = set()
     cited_dois = set()
     enriched_dois = set()
-    writers = {partition: ShardWriter(output, partition, args.rows_per_shard)
-               for partition in sorted(set(PARTITIONS.values()))}
-    examples = {partition: [] for partition in writers}
+    writer = ShardWriter(output, args.rows_per_shard)
+    examples = []
 
     for item in inputs:
         parquet = pq.ParquetFile(args.source_dir / item["path"])
@@ -181,10 +175,6 @@ def main():
                 counts["source_citing_entries"] += len(row["cited_by"])
                 counts["source_reference_matches"] += row["reference_count"]
                 for citing in row["cited_by"]:
-                    provenance = citing["provenance"]
-                    if provenance not in PARTITIONS:
-                        raise ValueError(f"{doi}: unknown citing provenance {provenance!r}")
-                    partition = PARTITIONS[provenance]
                     value = {"relatedIdentifier": require_doi(citing["doi"], "citing"),
                              "relatedIdentifierType": "DOI",
                              "relationType": "IsCitedBy"}
@@ -200,20 +190,18 @@ def main():
                               "field": "relatedIdentifiers", "enrichedValue": value,
                               "sourceId": args.source_id, "contentKey": key}
                     validate_enrichment(record)
-                    writers[partition].write(json.dumps(record, ensure_ascii=False,
-                                                        separators=(",", ":")) + "\n")
+                    writer.write(json.dumps(record, ensure_ascii=False,
+                                            separators=(",", ":")) + "\n")
                     counts["enrichments"] += 1
-                    by_partition[partition] += 1
-                    by_provenance[provenance] += 1
+                    by_provenance[citing["provenance"]] += 1
                     counts["enrichments_to_arxiv_dois"] += doi.startswith("10.48550/")
                     enriched_dois.add(doi)
-                    if len(examples[partition]) < 4:
-                        examples[partition].append(record)
+                    if len(examples) < 4:
+                        examples.append(record)
         print(json.dumps({"source_file": item["path"],
                           "source_rows": counts["source_rows"],
                           "enrichments": counts["enrichments"]}), flush=True)
-    for writer in writers.values():
-        writer.close()
+    writer.close()
 
     counts["unique_cited_dois"] = len(cited_dois)
     counts["unique_enriched_dois"] = len(enriched_dois)
@@ -222,7 +210,6 @@ def main():
     assert counts["source_rows"] == published["all"]["works"]
     assert counts["source_citing_entries"] == published["all"]["citations"]
     assert counts["source_reference_matches"] == published["all"]["references"]
-    assert by_partition["asserted"] == published["asserted"]["citations"]
     manifest = {
         "created_at": datetime.now(timezone.utc).isoformat(),
         "method": {"name": METHOD, "version": "1.0.0"},
@@ -238,14 +225,12 @@ def main():
                               "sha256": SCHEMA_SHA256},
         "content_key": {"algorithm": "xxh3-128", "canonicalization": "RFC 8785 JCS",
                         "input": "[method, doi, field, action, enrichedValue]"},
-        "partition_by_source_provenance": PARTITIONS,
         "counts": dict(sorted(counts.items())) | {
-            "enrichments_by_partition": dict(sorted(by_partition.items())),
             "enrichments_by_source_provenance": dict(sorted(by_provenance.items())),
         },
         "validation": {"enrichment_schema_failures": 0, "related_identifier_schema_failures": 0},
         "input_files": inputs,
-        "data_files": {partition: writer.files for partition, writer in writers.items()},
+        "data_files": writer.files,
     }
     (output / "manifest.json").write_text(json.dumps(manifest, indent=2) + "\n")
     (output / "provenance/examples.json").write_text(

manifest.json

@@ -1,5 +1,5 @@
 {
-  "created_at": "2026-10-14T13:21:01.789179+00:00",
+  "created_at": "2026-10-14T14:48:51.388656+00:00",
   "method": {
     "name": "crossref-datacite-citations",
     "version": "1.0.0"
@@ -25,11 +25,6 @@
     "canonicalization": "RFC 8785 JCS",
     "input": "[method, doi, field, action, enrichedValue]"
   },
-  "partition_by_source_provenance": {
-    "publisher": "asserted",
-    "crossref": "asserted",
-    "mined": "mined"
-  },
   "counts": {
     "enrichments": 7339071,
     "enrichments_to_arxiv_dois": 5999691,
@@ -38,10 +33,6 @@
     "source_rows": 1705826,
     "unique_cited_dois": 1705826,
     "unique_enriched_dois": 1705826,
-    "enrichments_by_partition": {
-      "asserted": 1269805,
-      "mined": 6069266
-    },
     "enrichments_by_source_provenance": {
       "crossref": 77601,
       "mined": 6069266,
@@ -72,196 +63,186 @@
       "sha256": "37691d142b900a377c05c144210829c033c2dc40831cc0cc2a84402e286e273f"
     }
   ],
-  "data_files": {
-    "asserted": [
-      {
-        "path": "data/asserted/part_0000.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7764452,
-        "sha256": "433c1c778bcdc86c96f26492cd6a65ac7a4bdf732b9fe4f260c553986c2fa766"
-      },
-      {
-        "path": "data/asserted/part_0001.jsonl.gz",
-        "rows": 250000,
-        "bytes": 8512251,
-        "sha256": "26d8170b2a3c9fd6350d969493096573f7493cac922ca53e0880e40b3b552d12"
-      },
-      {
-        "path": "data/asserted/part_0002.jsonl.gz",
-        "rows": 250000,
-        "bytes": 8981642,
-        "sha256": "cadd4cc436bc66bc0a9fbc0604ecae2f89e5400734e476cb44bb985289f7206e"
-      },
-      {
-        "path": "data/asserted/part_0003.jsonl.gz",
-        "rows": 250000,
-        "bytes": 9278585,
-        "sha256": "ac85547ec0dfaa4f31ae10684d7c858f94258749c631ca61964c927aa42e35b2"
-      },
-      {
-        "path": "data/asserted/part_0004.jsonl.gz",
-        "rows": 250000,
-        "bytes": 9174239,
-        "sha256": "e083301f1d5d4299ed160afd6d9b81a14e6efc164ede263bc9839f091685c7e7"
-      },
-      {
-        "path": "data/asserted/part_0005.jsonl.gz",
-        "rows": 19805,
-        "bytes": 731580,
-        "sha256": "6b863ddd303070ae4039d0566b7e86b560c37d817e684fb9749ffb1f7d7a557b"
-      }
-    ],
-    "mined": [
-      {
-        "path": "data/mined/part_0000.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7138618,
-        "sha256": "df9549fa79e49924b1bd207f30323dde7254dc487bd2dd124766bf2e6c086acf"
-      },
-      {
-        "path": "data/mined/part_0001.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7423605,
-        "sha256": "bfa2d00298b9e94ac94c0c6ba78f4cf6d774bd48938d9400e39db7fee0bb0a48"
-      },
-      {
-        "path": "data/mined/part_0002.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7583792,
-        "sha256": "896a2b337646287b80b599d6a912eff0f8eca2fee99ac6e629e161ddb2b126b2"
-      },
-      {
-        "path": "data/mined/part_0003.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7579789,
-        "sha256": "7b90253c07de6cacfbfb1df56b58183af90a0b147abc53ec6a18d746505e7205"
-      },
-      {
-        "path": "data/mined/part_0004.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7646179,
-        "sha256": "803d0f654f0dbaefcf03b3ce4fc973532f86314bf42810634a7af1723cda510c"
-      },
-      {
-        "path": "data/mined/part_0005.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7608419,
-        "sha256": "06a5de9e7bfe5ec3715264d87b0094de5ba5fc137cfb3a0017ea9ca84695abde"
-      },
-      {
-        "path": "data/mined/part_0006.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7632894,
-        "sha256": "a66bf3bf1caf7e1562e284da9d6ada76961999de48fbf7874af1b37d41c488e5"
-      },
-      {
-        "path": "data/mined/part_0007.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7674830,
-        "sha256": "e1bbbdaed0bd21e67e116afabe512168f6795a689ca142fa5beef576fc9f2801"
-      },
-      {
-        "path": "data/mined/part_0008.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7706750,
-        "sha256": "02bb274c59df4f9ef574f69bb6b5dd4c0c04bbfdffb75c4e6245fb327e969f1c"
-      },
-      {
-        "path": "data/mined/part_0009.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7764049,
-        "sha256": "d0baad39b54517f2d74b5f9ccf713b8d3554dcfc1280881d4f04440581263f6a"
-      },
-      {
-        "path": "data/mined/part_0010.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7836755,
-        "sha256": "7f15a475318189ad4d95159ecb6244ac146c05f44648545d1285794bdc15ef19"
-      },
-      {
-        "path": "data/mined/part_0011.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7917579,
-        "sha256": "bf833b0aabbe21d0216231c70cc733fe2167cb785229d1d299309b7b1354c5bd"
-      },
-      {
-        "path": "data/mined/part_0012.jsonl.gz",
-        "rows": 250000,
-        "bytes": 7999097,
-        "sha256": "bfff38279a0deb73c53b2f2c9e4faeebf4b498f7f5eecb106dbd6f7d5e5465e9"
-      },
-      {
-        "path": "data/mined/part_0013.jsonl.gz",
-        "rows": 250000,
-        "bytes": 8105868,
-        "sha256": "26c83e6a3dff4a02d4497ac11067514899a970365b7abf3f07d968048c46b367"
-      },
-      {
-        "path": "data/mined/part_0014.jsonl.gz",
-        "rows": 250000,
-        "bytes": 8196029,
-        "sha256": "ad32b5ce9da725c9329a13e68e6b4faf7e806a1064e468f3ecd8a2221206365f"
-      },
-      {
-        "path": "data/mined/part_0015.jsonl.gz",
-        "rows": 250000,
-        "bytes": 8326023,
-        "sha256": "91e900f7b3854b90927bb880f87e0b87f4e3036fd605f85927879eba36f259dc"
-      },
-      {
-        "path": "data/mined/part_0016.jsonl.gz",
-        "rows": 250000,
-        "bytes": 8438971,
-        "sha256": "83cc2b54ab0ef2f31d2f0de69ba873ad5b029e3fd516f9957a23b7207b091066"
-      },
-      {
-        "path": "data/mined/part_0017.jsonl.gz",
-        "rows": 250000,
-        "bytes": 8587743,
-        "sha256": "a332e903a300cc04288aa41b849d35b87790cd25605243418d30f8ca17239031"
-      },
-      {
-        "path": "data/mined/part_0018.jsonl.gz",
-        "rows": 250000,
-        "bytes": 8703475,
-        "sha256": "bab54ddbb4db605201f686c2f89f6ae8481841f5064f72a789324b959fecc972"
-      },
-      {
-        "path": "data/mined/part_0019.jsonl.gz",
-        "rows": 250000,
-        "bytes": 8743565,
-        "sha256": "56b63ced083f309cf2f5ef74cf6dcc3f93a47080e4e2a5a9f19200164f66ab74"
-      },
-      {
-        "path": "data/mined/part_0020.jsonl.gz",
-        "rows": 250000,
-        "bytes": 8998239,
-        "sha256": "0945365b7117e8f72a8454b0166c2a24cd7d2cb5eee32c0dfaedb553ecd93043"
-      },
-      {
-        "path": "data/mined/part_0021.jsonl.gz",
-        "rows": 250000,
-        "bytes": 9276102,
-        "sha256": "f169bfb26edeed932833dbed5439fdc6bfe4c962e96e60232a298912cb9bf66d"
-      },
-      {
-        "path": "data/mined/part_0022.jsonl.gz",
-        "rows": 250000,
-        "bytes": 9162521,
-        "sha256": "d458c77596b2bfb4faf4a4a379c0ec54fb37a86c677b920b4391383c3cb6fa61"
-      },
-      {
-        "path": "data/mined/part_0023.jsonl.gz",
-        "rows": 250000,
-        "bytes": 9052383,
-        "sha256": "bb192d61a4806b5a366d8f843a2f236210d4e500f117f469862c1ae1e8531259"
-      },
-      {
-        "path": "data/mined/part_0024.jsonl.gz",
-        "rows": 69266,
-        "bytes": 2602621,
-        "sha256": "f70eb015fea9294da0f06d10bf8f1c5f4395e503b0f0e095f6598985464a35a7"
-      }
-    ]
-  }
+  "data_files": [
+    {
+      "path": "data/part_0000.jsonl.gz",
+      "rows": 250000,
+      "bytes": 7053965,
+      "sha256": "cfba612307af500d2fb9ff6770cc2322bdc2de446cde21db760db774a87c7c37"
+    },
+    {
+      "path": "data/part_0001.jsonl.gz",
+      "rows": 250000,
+      "bytes": 7352024,
+      "sha256": "f6c7d160186f49930a7a3de254ed0d7ab2bfc9b2f06463b804045195d8117542"
+    },
+    {
+      "path": "data/part_0002.jsonl.gz",
+      "rows": 250000,
+      "bytes": 7529839,
+      "sha256": "f3b018faad9d4cf8f881c280d859c9c8228e3893cf6b9e086a74ca24a55bba9f"
+    },
+    {
+      "path": "data/part_0003.jsonl.gz",
+      "rows": 250000,
+      "bytes": 7614354,
+      "sha256": "44131976d8d69d7dedc498f13c8527f44b16514ad210d404ed43a21b8bfcc801"
+    },
+    {
+      "path": "data/part_0004.jsonl.gz",
+      "rows": 250000,
+      "bytes": 7631869,
+      "sha256": "4d72e4928bc9eef316dc631ce88ba253d24b7c7c44a31f64dc46ffc65d69d609"
+    },
+    {
+      "path": "data/part_0005.jsonl.gz",
+      "rows": 250000,
+      "bytes": 7680352,
+      "sha256": "78dd7a8c01e4d0e983c862c1d8fcf92e110db3b19aca4bd5678730b50e10f0db"
+    },
+    {
+      "path": "data/part_0006.jsonl.gz",
+      "rows": 250000,
+      "bytes": 7655637,
+      "sha256": "6546227319ea88df8fa10553450bed1e0e9eb9aa2e739d309dae4ccfc2157c84"
+    },
+    {
+      "path": "data/part_0007.jsonl.gz",
+      "rows": 250000,
+      "bytes": 7672781,
+      "sha256": "d232532165a6aa842306245f0683b0f015a5ec60d435797e4c4375042cddb02e"
+    },
+    {
+      "path": "data/part_0008.jsonl.gz",
+      "rows": 250000,
+      "bytes": 7727281,
+      "sha256": "2e72f2271f46168da1079a4e3eae2b69387a1fbf17285a49788017f477b544dc"
+    },
+    {
+      "path": "data/part_0009.jsonl.gz",
+      "rows": 250000,
+      "bytes": 7743195,
+      "sha256": "8042b37c82569325dedea7555069f8426725b84cec6309d2256daa4265a36bea"
+    },
+    {
+      "path": "data/part_0010.jsonl.gz",
+      "rows": 250000,
+      "bytes": 7781733,
+      "sha256": "f006f9b4e89c7468c84b2bd558cb34561efcedc305a37dab6170a137cce7c518"
+    },
+    {
+      "path": "data/part_0011.jsonl.gz",
+      "rows": 250000,
+      "bytes": 7859045,
+      "sha256": "58492349436e9e8521655422b61922299a537265acff3dd6827442148ac68334"
+    },
+    {
+      "path": "data/part_0012.jsonl.gz",
+      "rows": 250000,
+      "bytes": 7918318,
+      "sha256": "7d7ef0b9b2ca436be60ac83d03a432a3c73f0f5bf03f171c8ea933b8cff51eb8"
+    },
+    {
+      "path": "data/part_0013.jsonl.gz",
+      "rows": 250000,
+      "bytes": 8002799,
+      "sha256": "675788210e89fd7aa1c607ccdddfd8ffbc32e32dd56db5ca17cdc9359464b12d"
+    },
+    {
+      "path": "data/part_0014.jsonl.gz",
+      "rows": 250000,
+      "bytes": 8063594,
+      "sha256": "fe571c107e17f5711dabea6dee99fd531cdb0da320dafbf0e88b8efb43494cd0"
+    },
+    {
+      "path": "data/part_0015.jsonl.gz",
+      "rows": 250000,
+      "bytes": 8171822,
+      "sha256": "195d3599253d715b21b2997d9e5c60f04ab640e70199d474503d64f79db84da1"
+    },
+    {
+      "path": "data/part_0016.jsonl.gz",
+      "rows": 250000,
+      "bytes": 8282712,
+      "sha256": "4a3d33118cbdd93920e338eb610ee41f6e7589ea93c3799d3b6e756db8e93c4a"
+    },
+    {
+      "path": "data/part_0017.jsonl.gz",
+      "rows": 250000,
+      "bytes": 8330250,
+      "sha256": "4e61bf2478bb37178fa5750bfc83bdb5449f13d691727ed8b3117c2029c7971e"
+    },
+    {
+      "path": "data/part_0018.jsonl.gz",
+      "rows": 250000,
+      "bytes": 8459090,
+      "sha256": "88b1e6072766b8a0772adf951fe121d980a2cc069fe2d04cfff861802e15445f"
+    },
+    {
+      "path": "data/part_0019.jsonl.gz",
+      "rows": 250000,
+      "bytes": 8543045,
+      "sha256": "ad3a7d8cbefa1dd472199773691c9139eb97a57ebfd792fcbedfc8bf79d5843b"
+    },
+    {
+      "path": "data/part_0020.jsonl.gz",
+      "rows": 250000,
+      "bytes": 8650606,
+      "sha256": "89d14ad3cd0d8c70aee0eb6e30d8b84d01500e45f5c70c49129b3d2a2b7b077a"
+    },
+    {
+      "path": "data/part_0021.jsonl.gz",
+      "rows": 250000,
+      "bytes": 8799712,
+      "sha256": "c1cf9d163ffed7efcad38211507ebd4c01e51243b0cba370718cac72b3c4b0e7"
+    },
+    {
+      "path": "data/part_0022.jsonl.gz",
+      "rows": 250000,
+      "bytes": 8746213,
+      "sha256": "45182b3fce835539a3efc59677a8d500c392fbe31e39b0db0a2d27c0a5bf4c9e"
+    },
+    {
+      "path": "data/part_0023.jsonl.gz",
+      "rows": 250000,
+      "bytes": 9013941,
+      "sha256": "2e43db94e60f5537d83cf6a4815ded2254d3847fdd7df1aef5f0d9420bcea1a8"
+    },
+    {
+      "path": "data/part_0024.jsonl.gz",
+      "rows": 250000,
+      "bytes": 8928359,
+      "sha256": "d94de5a2d08f23cb5ad7e708745bc69113bd305e53ec9e9a5e4e02cdc64f5fed"
+    },
+    {
+      "path": "data/part_0025.jsonl.gz",
+      "rows": 250000,
+      "bytes": 9383984,
+      "sha256": "007f363d7e0d224c6b2248cfc41e410db91c365c0e1e5fd2b8acb4aac92b06ad"
+    },
+    {
+      "path": "data/part_0026.jsonl.gz",
+      "rows": 250000,
+      "bytes": 9389960,
+      "sha256": "58be99910aededee2ef7c70673926f4fb45f624584796dcb36dd2c74538db83c"
+    },
+    {
+      "path": "data/part_0027.jsonl.gz",
+      "rows": 250000,
+      "bytes": 9187332,
+      "sha256": "c4cf5137c5b8c6bf9776b4273dba4434935b29aede40735633ef15b422af55fc"
+    },
+    {
+      "path": "data/part_0028.jsonl.gz",
+      "rows": 250000,
+      "bytes": 9079174,
+      "sha256": "0f87a91a2db24b7a107806b2dc3c85b5d4c2b9a9a9cc3f24874e17e199ddec84"
+    },
+    {
+      "path": "data/part_0029.jsonl.gz",
+      "rows": 89071,
+      "bytes": 3241535,
+      "sha256": "74e99f3262658a1017522af5f6a45af5d95bfe822ba364ee48f2680fcc9c0d48"
+    }
+  ]
 }

provenance/examples.json

@@ -1,102 +1,50 @@
-{
-  "asserted": [
-    {
-      "doi": "10.48550/arxiv.1412.6980",
-      "action": "insert",
-      "field": "relatedIdentifiers",
-      "enrichedValue": {
-        "relatedIdentifier": "10.1002/9781394372539.ch6",
-        "relatedIdentifierType": "DOI",
-        "relationType": "IsCitedBy"
-      },
-      "sourceId": "",
-      "contentKey": "8f221c7f80f2b7bad183b55359e24bb7"
+[
+  {
+    "doi": "10.48550/arxiv.1412.6980",
+    "action": "insert",
+    "field": "relatedIdentifiers",
+    "enrichedValue": {
+      "relatedIdentifier": "10.1001/jamadermatol.2022.5415",
+      "relatedIdentifierType": "DOI",
+      "relationType": "IsCitedBy"
     },
-    {
-      "doi": "10.48550/arxiv.1412.6980",
-      "action": "insert",
-      "field": "relatedIdentifiers",
-      "enrichedValue": {
-        "relatedIdentifier": "10.1002/acm2.13951",
-        "relatedIdentifierType": "DOI",
-        "relationType": "IsCitedBy"
-      },
-      "sourceId": "",
-      "contentKey": "c2d39280df91ed94f73519156a2aa1da"
+    "sourceId": "",
+    "contentKey": "63f52c05c46e67981b4f1f5157606c91"
+  },
+  {
+    "doi": "10.48550/arxiv.1412.6980",
+    "action": "insert",
+    "field": "relatedIdentifiers",
+    "enrichedValue": {
+      "relatedIdentifier": "10.1001/jamanetworkopen.2018.6937",
+      "relatedIdentifierType": "DOI",
+      "relationType": "IsCitedBy"
     },
-    {
-      "doi": "10.48550/arxiv.1412.6980",
-      "action": "insert",
-      "field": "relatedIdentifiers",
-      "enrichedValue": {
-        "relatedIdentifier": "10.1002/acm2.13978",
-        "relatedIdentifierType": "DOI",
-        "relationType": "IsCitedBy"
-      },
-      "sourceId": "",
-      "contentKey": "9768047ede9f90352cc1c155afbacb60"
+    "sourceId": "",
+    "contentKey": "67789564734ccd529749cf9b05ccefd7"
+  },
+  {
+    "doi": "10.48550/arxiv.1412.6980",
+    "action": "insert",
+    "field": "relatedIdentifiers",
+    "enrichedValue": {
+      "relatedIdentifier": "10.1001/jamanetworkopen.2019.3963",
+      "relatedIdentifierType": "DOI",
+      "relationType": "IsCitedBy"
     },
-    {
-      "doi": "10.48550/arxiv.1412.6980",
-      "action": "insert",
-      "field": "relatedIdentifiers",
-      "enrichedValue": {
-        "relatedIdentifier": "10.1002/adma.202514150",
-        "relatedIdentifierType": "DOI",
-        "relationType": "IsCitedBy"
-      },
-      "sourceId": "",
-      "contentKey": "77f475a12788ec5116f98a71c5790eca"
-    }
-  ],
-  "mined": [
-    {
-      "doi": "10.48550/arxiv.1412.6980",
-      "action": "insert",
-      "field": "relatedIdentifiers",
-      "enrichedValue": {
-        "relatedIdentifier": "10.1001/jamadermatol.2022.5415",
-        "relatedIdentifierType": "DOI",
-        "relationType": "IsCitedBy"
-      },
-      "sourceId": "",
-      "contentKey": "63f52c05c46e67981b4f1f5157606c91"
+    "sourceId": "",
+    "contentKey": "697ad2aed03f2716ae894f669151b26e"
+  },
+  {
+    "doi": "10.48550/arxiv.1412.6980",
+    "action": "insert",
+    "field": "relatedIdentifiers",
+    "enrichedValue": {
+      "relatedIdentifier": "10.1001/jamanetworkopen.2020.15626",
+      "relatedIdentifierType": "DOI",
+      "relationType": "IsCitedBy"
     },
-    {
-      "doi": "10.48550/arxiv.1412.6980",
-      "action": "insert",
-      "field": "relatedIdentifiers",
-      "enrichedValue": {
-        "relatedIdentifier": "10.1001/jamanetworkopen.2018.6937",
-        "relatedIdentifierType": "DOI",
-        "relationType": "IsCitedBy"
-      },
-      "sourceId": "",
-      "contentKey": "67789564734ccd529749cf9b05ccefd7"
-    },
-    {
-      "doi": "10.48550/arxiv.1412.6980",
-      "action": "insert",
-      "field": "relatedIdentifiers",
-      "enrichedValue": {
-        "relatedIdentifier": "10.1001/jamanetworkopen.2019.3963",
-        "relatedIdentifierType": "DOI",
-        "relationType": "IsCitedBy"
-      },
-      "sourceId": "",
-      "contentKey": "697ad2aed03f2716ae894f669151b26e"
-    },
-    {
-      "doi": "10.48550/arxiv.1412.6980",
-      "action": "insert",
-      "field": "relatedIdentifiers",
-      "enrichedValue": {
-        "relatedIdentifier": "10.1001/jamanetworkopen.2020.15626",
-        "relatedIdentifierType": "DOI",
-        "relationType": "IsCitedBy"
-      },
-      "sourceId": "",
-      "contentKey": "c92e667f712992469f9138feed793841"
-    }
-  ]
-}
+    "sourceId": "",
+    "contentKey": "c92e667f712992469f9138feed793841"
+  }
+]

provenance/independent_projection.sql

@@ -1,13 +1,2 @@
-WITH pairs AS (
-  SELECT s.doi, cb.doi related_identifier,
-         list_transform(cb.matches, m -> m.provenance) match_provenances
-  FROM source s, unnest(s.cited_by) AS t(cb)
-)
-SELECT
-  doi,
-  related_identifier,
-  CASE
-    WHEN list_has_any(match_provenances, ['publisher', 'crossref']) THEN 'asserted'
-    WHEN list_contains(match_provenances, 'mined') THEN 'mined'
-  END AS partition
-FROM pairs
+SELECT s.doi, cb.doi AS related_identifier
+FROM source s, unnest(s.cited_by) AS t(cb)

validate.py

@@ -32,16 +32,15 @@ def main():
         (root / "provenance/related_identifier_subset_schema.json").read_text()))
     expected = manifest["counts"]
     counts = Counter()
-    by_partition = Counter()
     keys = set()
     dois = set()
-    files = [(partition, item) for partition, items in manifest["data_files"].items()
-             for item in items]
+    files = manifest["data_files"]
+    # Every file under data/, not only data/*.jsonl.gz, so a leftover
+    # subdirectory that the default config would not load also fails.
     on_disk = {str(p.relative_to(root)) for p in (root / "data").rglob("*") if p.is_file()}
-    listed = {item["path"] for _, item in files}
+    listed = {item["path"] for item in files}
     assert on_disk == listed, f"unlisted {sorted(on_disk - listed)}, missing {sorted(listed - on_disk)}"
-    for partition, item in files:
-        assert Path(item["path"]).parent == Path("data", partition), f"{item['path']} listed as {partition}"
+    for item in files:
         path = root / item["path"]
         with path.open("rb") as stream:
             digest = hashlib.file_digest(stream, "sha256").hexdigest()
@@ -74,17 +73,14 @@ def main():
                 dois.add(row["doi"])
                 counts["enrichments"] += 1
                 counts["to_arxiv_dois"] += row["doi"].startswith("10.48550/")
-                by_partition[partition] += 1
                 n += 1
         assert n == item["rows"], f"{item['path']}: {n} records != manifest {item['rows']}"
         print(f"Validated {item['path']}: {n:,} records", flush=True)
     assert counts["enrichments"] == expected["enrichments"]
     assert len(dois) == expected["unique_enriched_dois"]
     assert counts["to_arxiv_dois"] == expected["enrichments_to_arxiv_dois"]
-    assert dict(by_partition) == expected["enrichments_by_partition"]
     report = {"status": "passed", "records_read_and_validated": counts["enrichments"],
               "unique_target_dois": len(dois), "data_shards": len(files),
-              "enrichments_by_partition": dict(sorted(by_partition.items())),
               "enrichments_to_arxiv_dois": counts["to_arxiv_dois"],
               "schema_errors": 0, "content_key_mismatches": 0, "duplicate_content_keys": 0,
               "file_checksum_mismatches": 0}

validation_report.json

@@ -2,11 +2,7 @@
   "status": "passed",
   "records_read_and_validated": 7339071,
   "unique_target_dois": 1705826,
-  "data_shards": 31,
-  "enrichments_by_partition": {
-    "asserted": 1269805,
-    "mined": 6069266
-  },
+  "data_shards": 30,
   "enrichments_to_arxiv_dois": 5999691,
   "schema_errors": 0,
   "content_key_mismatches": 0,
@@ -18,10 +14,6 @@
     "actual_citations": 7339071,
     "missing_citations": 0,
     "extra_citations": 0,
-    "actual_citations_by_partition": {
-      "asserted": 1269805,
-      "mined": 6069266
-    },
     "records_with_unexpected_constants": 0
   }
 }

verify_source.py

@@ -1,5 +1,5 @@
 #!/usr/bin/env python3
-"""Compare every emitted citation tuple with an independent SQL source projection."""
+"""Compare every emitted citation pair with an independent SQL source projection."""
 import argparse
 import json
 from pathlib import Path
@@ -22,26 +22,22 @@ def main():
     c.read_parquet(str(args.source_dir / "*.parquet")).create_view("source")
     sql = (args.dataset_dir / "provenance/independent_projection.sql").read_text()
     c.execute("CREATE TEMP TABLE expected AS " + sql)
-    c.read_json(str(args.dataset_dir / "data/*/*.jsonl.gz"), format="newline_delimited",
-                filename=True).create_view("records")
+    c.read_json(str(args.dataset_dir / "data/*.jsonl.gz"),
+                format="newline_delimited").create_view("records")
     unexpected_constants = c.execute("""SELECT count(*) FROM records
         WHERE enrichedValue.relatedIdentifierType IS DISTINCT FROM 'DOI'
            OR enrichedValue.relationType IS DISTINCT FROM 'IsCitedBy'""").fetchone()[0]
     c.execute("""CREATE TEMP TABLE actual AS
-        SELECT doi, enrichedValue.relatedIdentifier AS related_identifier,
-               string_split(filename, '/')[-2] AS partition
-        FROM records""")
+        SELECT doi, enrichedValue.relatedIdentifier AS related_identifier FROM records""")
     missing = c.execute("SELECT count(*) FROM (SELECT * FROM expected EXCEPT SELECT * FROM actual)").fetchone()[0]
     extra = c.execute("SELECT count(*) FROM (SELECT * FROM actual EXCEPT SELECT * FROM expected)").fetchone()[0]
     expected_count = c.execute("SELECT count(*) FROM expected").fetchone()[0]
     actual_count = c.execute("SELECT count(*) FROM actual").fetchone()[0]
-    by_partition = dict(c.execute(
-        "SELECT partition, count(*) FROM actual GROUP BY ALL ORDER BY ALL").fetchall())
     passed = missing == extra == unexpected_constants == 0 and expected_count == actual_count
     result = {"status": "passed" if passed else "failed",
               "independent_expected_citations": expected_count,
               "actual_citations": actual_count, "missing_citations": missing,
-              "extra_citations": extra, "actual_citations_by_partition": by_partition,
+              "extra_citations": extra,
               "records_with_unexpected_constants": unexpected_constants}
     print(json.dumps(result, indent=2))
     assert passed