Replace dataset with 41.6M-pipeline output (UC fixes) — phase 1

Phase 1 of UC-feedback regeneration:
- New README with full ground-truth, scoring, and MITRE documentation
- Delete stale signal shards (21 new shards uploaded in phase 2)
- Delete placeholder graph.graphml + graph.json (replaced by per-incident GraphMLs)
- New nodes.jsonl (44,717 nodes), attack_reports.jsonl (12,361 reports)

README.md

@@ -18,12 +18,12 @@ tags:
   - labeled-dataset
   - lead-rules
 size_categories:
-  - 10M<n<100M
+  - 100M<n<1B
 dataset_info:
   - config_name: signals
     splits:
       - name: train
-        num_examples: 41658748
+        num_examples: 114234041
 configs:
   - config_name: signals
     data_files:
@@ -47,11 +47,11 @@ configs:
 
 ## Overview
 
-A large-scale, labeled cybersecurity dataset derived from production Security Operations Center (SOC) data processed by [WitFoo Precinct](https://www.witfoo.com/) version 6.x. This dataset contains **41,658,748 sanitized security events** (signal logs) across 5 organizations and **12,361 incident provenance graphs** (44,717 nodes, 9,494,858 edges).
+A large-scale, labeled cybersecurity dataset derived from production Security Operations Center (SOC) data processed by [WitFoo Precinct](https://www.witfoo.com/) version 6.x. This dataset contains **114,234,041 sanitized security events** (signal logs) across 5 organizations and **12,361 incident provenance graphs** (47,632 nodes, 32,086,552 edges).
 
 **Available in two sizes:**
 - [`witfoo/precinct6-cybersecurity`](https://huggingface.co/datasets/witfoo/precinct6-cybersecurity) — 2.1M signals (smaller, faster to load; same incidents and graph methodology)
-- [`witfoo/precinct6-cybersecurity-100m`](https://huggingface.co/datasets/witfoo/precinct6-cybersecurity-100m) — **41.6M signals (this dataset)**
+- [`witfoo/precinct6-cybersecurity-100m`](https://huggingface.co/datasets/witfoo/precinct6-cybersecurity-100m) — **114M signals (this dataset)**
 
 **Generate your own:** WitFoo Precinct 6.x customers can create datasets from their own data using the open-source pipeline: [`witfoo/dataset-from-precinct6`](https://github.com/witfoo/dataset-from-precinct6)
 
@@ -67,7 +67,7 @@ This dataset is designed to support research in:
 ```python
 from datasets import load_dataset
 
-# Load flat signal logs (41.6M rows across 21 Parquet shards)
+# Load flat signal logs (114M rows across 58 Parquet shards)
 signals = load_dataset("witfoo/precinct6-cybersecurity-100m", "signals", split="train")
 
 # Find analyst-confirmed malicious events
@@ -78,7 +78,7 @@ malicious_confirmed = signals.filter(
 # Find suspicious events that matched detection rules but are not in confirmed incidents
 suspicious = signals.filter(lambda x: x["label_binary"] == "suspicious")
 
-# Load provenance graph (44.7k nodes, 9.5M edges)
+# Load provenance graph (47.6k nodes, 32M edges)
 nodes = load_dataset("witfoo/precinct6-cybersecurity-100m", "graph_nodes", split="train")
 edges = load_dataset("witfoo/precinct6-cybersecurity-100m", "graph_edges", split="train")
 
@@ -90,17 +90,17 @@ incidents = load_dataset("witfoo/precinct6-cybersecurity-100m", "incidents", spl
 
 | Label | Count | Percentage |
 |-------|-------|------------|
-| `benign` | 41,447,721 | 99.49% |
-| `malicious` | 148,128 | 0.36% |
-| `suspicious` | 62,899 | 0.15% |
+| `benign` | 113,543,372 | 99.40% |
+| `suspicious` | 616,605 | 0.54% |
+| `malicious` | 74,064 | 0.06% |
 
 Disposition breakdown (the raw Precinct status, exposed for ground-truth stratification — see [Ground Truth](#ground-truth-and-disposition)):
 
 | Disposition | Count | Meaning |
 |-------------|-------|---------|
-| `Unprocessed` | 41,575,946 | No analyst review (default state for benign/suspicious + un-reviewed malicious) |
-| `Disrupted` | 82,782 | SOC analyst confirmed and intervened |
-| `Dismissed` | 20 | SOC analyst dismissed |
+| `Unprocessed` | 114,192,640 | No analyst review (default state for benign/suspicious + un-reviewed malicious) |
+| `Disrupted` | 41,391 | SOC analyst confirmed and intervened |
+| `Dismissed` | 10 | SOC analyst dismissed |
 
 ## Signal Columns
 
@@ -159,9 +159,9 @@ The dataset contains events from **158 security products** across **70+ vendors*
 
 **Three-tier labels** derived from two sources:
 
-- **`malicious`** (148,128): Events embedded as leads inside confirmed incidents. Extracted directly from incident lead objects with suspicion scores, modus operandi, and MITRE mappings.
-- **`suspicious`** (62,899): Events matching WitFoo's 261 lead detection rules but not present in confirmed incidents.
-- **`benign`** (41,447,721): Events not matching any detection rules and not in any incident.
+- **`malicious`** (74,064): Events embedded as leads inside confirmed incidents. Extracted directly from incident lead objects with suspicion scores, modus operandi, and MITRE mappings.
+- **`suspicious`** (616,605): Events matching WitFoo's 261 lead detection rules but not present in confirmed incidents.
+- **`benign`** (113,543,372): Events not matching any detection rules and not in any incident.
 
 ### Ground Truth and Disposition
 
@@ -176,7 +176,7 @@ The dataset contains events from **158 security products** across **70+ vendors*
 | `False Positive` | SOC analyst confirmed false positive | Negative — analyst rejected |
 | `Unprocessed` | Automated detection, no human review | Lower — Precinct-confidence only |
 
-The `disposition_category` column buckets these into four values for easier filtering: `confirmed-malicious`, `false-positive`, `dismissed`, `automated`. For experiments where ground-truth quality matters, restrict to records where `disposition` ∈ {`Disrupted`, `Resolved`} to compare against analyst-confirmed labels. **This dataset has 82,782 records with `Disrupted` status, 20 with `Dismissed`, and the remainder with `Unprocessed` (no analyst review).**
+The `disposition_category` column buckets these into four values for easier filtering: `confirmed-malicious`, `false-positive`, `dismissed`, `automated`. For experiments where ground-truth quality matters, restrict to records where `disposition` ∈ {`Disrupted`, `Resolved`} to compare against analyst-confirmed labels. **This dataset has 41,391 records with `Disrupted` status, 10 with `Dismissed`, and the remainder with `Unprocessed` (no analyst review).**
 
 For benign and suspicious records, `disposition` is `Unprocessed` (no incident association). For malicious records, `disposition` reflects the parent incident's status at extraction time.
 
@@ -212,8 +212,8 @@ Tactic IDs use the standard MITRE ATT&CK Enterprise format (`TA0001` through `TA
 
 | Component | Count |
 |-----------|-------|
-| Nodes (artifact-derived hosts + credentials) | 44,717 |
-| Edges (artifact `EVENT`/`NETWORK_FLOW` + incident `INCIDENT_LINK`) | 9,494,858 |
+| Nodes (artifact-derived hosts + credentials) | 47,632 |
+| Edges (artifact `EVENT`/`NETWORK_FLOW` + incident `INCIDENT_LINK`) | 32,086,552 |
 | Incidents | 12,361 |
 
 Per-edge labels in `edges.jsonl` include `attack_techniques`, `attack_tactics`, `set_roles`, `lifecycle_stage`, `disposition`, `mo_name`, `suspicion_score`, `incident_id` (for INCIDENT_LINK edges).
@@ -228,16 +228,16 @@ Researchers can audit exactly how each sentence is derived by reading [`src/prec
 
 ## Additional Files
 
-- **`signals/signals-NNNNN.parquet`** — 21 Parquet shards (~280 MB each), 2M rows per shard except final shard. Total ~5.9 GB.
+- **`signals/signals-NNNNN.parquet`** — 58 Parquet shards (~270 MB each), 2M rows per shard except final shard. Total ~15.7 GB.
 - **`signals/metadata.json`** — Per-shard summary, label/disposition distributions, top message types and streams.
-- **`graph/nodes.jsonl`** — 44,717 graph nodes (hosts + credentials derived from artifact src/dst/username).
-- **`graph/edges.jsonl`** — 9,494,858 graph edges, each with `labels` dict carrying MITRE, disposition, set_roles, etc.
+- **`graph/nodes.jsonl`** — 47,632 graph nodes (hosts + credentials derived from artifact src/dst/username).
+- **`graph/edges.jsonl`** — 32,086,552 graph edges, each with `labels` dict carrying MITRE, disposition, set_roles, etc.
 - **`graph/incidents.jsonl`** — Full incident records (12,361 lines, ~638 MB) with embedded `nodes`, `edges`, `leads`, and framework mappings.
 - **`graph/incidents_graphml/{0-f}/{incident_id}.graphml`** — **12,361 per-incident GraphML files**, sharded into 16 subdirectories by first hex char of the incident UUID (HuggingFace caps directories at 10,000 files). Each file is small (KB-MB) and loadable in Gephi, NetworkX, igraph, or DGL. Ideal for graph-based research where loading the entire dataset is impractical.
 - **`graph/attack_reports.jsonl`** — Natural-language threat-hunting reports (12,361 reports, ~21 MB). See [Attack Reports](#attack-reports).
 - **`reference/lead_rules_catalog.json`** — Complete catalog of 261 WitFoo lead detection rules, 158 security products, 106 classification sets, and 216 stream-to-product mappings.
 
-**Note on the artifact-level graph:** This 41.6M-row extraction yielded fewer events than the previously-published 114M version — the difference reflects Cassandra coordinator behaviour during full-partition extraction. Methodology and content fields are otherwise identical to the 2M version.
+**Note on the artifact-level graph:** With 32M edges, the monolithic `graph.graphml` is intentionally **not** shipped at this scale — per-incident GraphMLs and the streaming `edges.jsonl` are the recommended entry points. Methodology and content fields are otherwise identical to the 2M version.
 
 ## Sanitization
 
@@ -258,7 +258,7 @@ This dataset was produced in collaboration with the University of Canterbury (Ne
 
 ## Limitations
 
-- **Label imbalance**: 99.49% benign reflects production SOC reality. Sampling strategies needed for balanced training.
+- **Label imbalance**: 99.40% benign reflects production SOC reality. Sampling strategies needed for balanced training.
 - **Temporal scope**: July–August 2024.
 - **Ground truth**: All labels derive from Precinct's automated correlation. Use the `disposition` column to stratify by analyst review level.
 - **Shared incidents**: The same 12,361 incidents appear in both the 2M and this dataset (incidents are stored separately from signal data; only the signal sample size differs).

signals/metadata.json

@@ -1,120 +1,120 @@
 {
-  "total_records": 41658748,
-  "n_shards": 21,
-  "total_size_bytes": 6285096828,
+  "total_records": 114234041,
+  "n_shards": 58,
+  "total_size_bytes": 16903246723,
   "label_distribution": {
-    "benign": 41447721,
-    "suspicious": 62899,
-    "malicious": 148128
+    "benign": 113543372,
+    "suspicious": 616605,
+    "malicious": 74064
   },
   "disposition_distribution": {
-    "Unprocessed": 41575946,
-    "Disrupted": 82782,
-    "Dismissed": 20
+    "Unprocessed": 114192640,
+    "Disrupted": 41391,
+    "Dismissed": 10
   },
   "top_message_types": {
-    "security_audit_event": 9229251,
+    "security_audit_event": 31771712,
+    "audit_log": 6347890,
+    "AssumeRole": 5975795,
+    "DescribeInstanceStatus": 5432873,
+    "unknown": 4877309,
+    "GetLogEvents": 4283853,
     "4662": 2812876,
+    "DescribeInstances": 2736657,
     "5156": 2432714,
-    "audit_log": 2118371,
+    "DescribeLoadBalancers": 2374064,
     "4658": 2024740,
-    "AssumeRole": 1746338,
-    "DescribeInstanceStatus": 1585360,
-    "unknown": 1559702,
-    "GetLogEvents": 1256453,
+    "ListClusters": 1711426,
+    "DescribeAccountAttributes": 1487651,
+    "management_message": 1291032,
+    "GetBucketAcl": 1026837,
     "4656": 1014502,
+    "ListFunctions20150331": 1013352,
     "4690": 1013158,
-    "DescribeInstances": 794197,
-    "DescribeLoadBalancers": 689914,
-    "ListClusters": 499015,
-    "systemd_event": 458497,
-    "DescribeAccountAttributes": 432371,
-    "GetBucketAcl": 307794,
-    "ListFunctions20150331": 301350,
-    "UpdateInstanceInformation": 285744,
-    "diagnostic_event": 240546,
-    "communication": 233061,
-    "DescribeVolumes": 232224,
-    "pam_event": 227289,
-    "DescribeClusters": 221341,
-    "GetBucketLocation": 218321,
-    "DescribeVpcs": 218177,
-    "GetBucketMetricsConfiguration": 210779,
-    "account_logon": 206951,
-    "DescribeTargetGroups": 206590,
-    "ListStacks": 202445,
-    "GetResources": 201427,
-    "account_logoff": 199416,
-    "DescribeSubnets": 190895,
-    "UpdateInstanceAssociationStatus": 185677,
-    "DescribeAlarmHistory": 184829,
-    "GetSendQuota": 178419,
-    "DescribeAccountLimits": 177594,
-    "4776": 170092,
-    "ListTopics": 168937,
-    "DescribeLimits": 162647,
-    "ListTables": 156659,
-    "access_log": 146831,
-    "5158": 143544,
-    "18453": 141318,
-    "DescribeAddresses": 140900,
-    "DescribeAutoScalingGroups": 139013,
-    "DescribeImages": 135742,
-    "GenerateDataKey": 135246,
-    "DescribeInternetGateways": 131041,
-    "auth_event": 126943
+    "UpdateInstanceInformation": 971754,
+    "DescribeVolumes": 791146,
+    "GetBucketLocation": 768284,
+    "DescribeClusters": 760994,
+    "DescribeVpcs": 743495,
+    "DescribeTargetGroups": 701104,
+    "java_error": 696559,
+    "GetBucketMetricsConfiguration": 692798,
+    "GetResources": 679887,
+    "ListStacks": 679521,
+    "DescribeSubnets": 646638,
+    "DescribeAlarmHistory": 632309,
+    "UpdateInstanceAssociationStatus": 627080,
+    "DescribeAccountLimits": 617531,
+    "GetSendQuota": 614677,
+    "ListTopics": 577029,
+    "DescribeLimits": 555857,
+    "ListTables": 534764,
+    "DescribeAutoScalingGroups": 484532,
+    "18453": 475412,
+    "DescribeAddresses": 475162,
+    "DescribeImages": 465047,
+    "GenerateDataKey": 464884,
+    "firewall_action": 464677,
+    "systemd_event": 463136,
+    "DescribeInternetGateways": 449468,
+    "DescribeSecurityGroups": 409111,
+    "DescribeLaunchConfigurations": 408698,
+    "DescribeFileSystems": 407067,
+    "ListEventSourceMappings20150331": 400999,
+    "GetRestApis": 400254,
+    "ListInstanceAssociations": 399538
   },
   "top_streams": {
-    "aws_cloudtrail_events": 16087719,
-    "microsoft-windows-security-auditing": 10481332,
-    "windows_security_audit": 9229251,
-    "aws_cloud_trail": 1994628,
-    "no_useful_info": 1503924,
-    "crond": 432131,
+    "aws_cloudtrail_events": 54942980,
+    "windows_security_audit": 31771712,
+    "microsoft-windows-security-auditing": 10985995,
+    "aws_cloud_trail": 6224147,
+    "no_useful_info": 4673405,
+    "vcenter": 1203516,
+    "java_stack_trace": 696451,
+    "mssql$paygsql01": 655415,
+    "cisco_asa": 465013,
+    "crond": 435510,
     "sshd": 248142,
     "filebeat_diagnostic": 240546,
     "network_communication": 233061,
     "pam": 227289,
-    "mssql$paygsql01": 193458,
+    "unknown": 203203,
+    "aws_vpc_flow_log": 192277,
+    "dnsmasq": 154223,
     "systemd_logind": 126943,
-    "cisco_asa": 99488,
     "linux_audit": 90726,
-    "aws_vpc_flow_log": 58162,
-    "unknown": 55441,
+    "barracuda_waf": 71122,
     "security": 53331,
     "cisco_os": 51551,
     "ad fs auditing": 40738,
     "symantec_sep": 36946,
-    "meraki": 36418,
+    "service control manager": 35325,
     "ad_audit_plus": 33017,
-    "systemd": 26286,
-    "service control manager": 23244,
-    "pan_firewall": 8298,
-    "microsoft-windows-distributedcom": 7754,
-    "Integrations Status": 7192,
-    "Crowdstrike Asset Inventory": 6994,
-    "DUO Asset Inventory": 3767,
+    "microsoft-windows-distributedcom": 26912,
+    "systemd": 26556,
+    "meraki": 18209,
+    "Crowdstrike Asset Inventory": 16605,
+    "Integrations Status": 12080,
+    "DUO Asset Inventory": 4902,
+    "pan_firewall": 4149,
     "esent": 3704,
+    "Crowdstrike Detection": 3530,
     "generic_log": 3379,
     "directory synchronization": 3121,
+    "Microsoft Graph Asset Inventory": 2200,
+    "microsoft-windows-security-spp": 2142,
     "schannel": 2137,
-    "microsoft-windows-security-spp": 2039,
-    "Crowdstrike Detection": 1155,
+    "barracuda_ess": 1805,
+    "trend micro_deep security manager": 1046,
+    "rsyslogd": 988,
+    "sqlagent$paygsql01": 878,
+    "apache_access_log": 821,
+    "symantec_dlp": 714,
     "mssqlserver": 529,
-    "trend micro_deep security manager": 353,
+    ".net runtime": 426,
     "sentinelone": 337,
-    "microsoft-windows-certificateservicesclient-autoenrollment": 293,
-    "meraki_l7_firewall": 292,
-    "sqlagent$paygsql01": 264,
-    "mssql$microsoft##wid": 251,
-    "symantec_dlp": 221,
-    "volsnap": 187,
-    "vss": 187,
-    "nutanix guest agent": 154,
-    "vssaudit": 146,
-    ".net runtime": 135,
-    "msiinstaller": 123,
-    "adsync": 106
+    "microsoft-windows-was": 308
   },
   "schema": [
     "timestamp",