Hugging Face's logo

Datasets:
zchen798
/
as_feature_snapshot

Dataset card Files
xet
 Community
Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
snapshot_month: string
snapshot_id: string
schema_version: string
generated_at_utc: string
artifacts: struct<snapshot_parquet: string, manifest: string, schema_json: string, schema_md: string, data_sour (... 12 chars omitted)
  child 0, snapshot_parquet: string
  child 1, manifest: string
  child 2, schema_json: string
  child 3, schema_md: string
  child 4, data_sources: string
snapshot: struct<format: string, compression: string, num_asns: int64, columns: list<item: string>, column_sto (... 2902 chars omitted)
  child 0, format: string
  child 1, compression: string
  child 2, num_asns: int64
  child 3, columns: list<item: string>
      child 0, item: string
  child 4, column_storage_types: struct<asn: string, LACeS-anycast_v4_cnt: string, LACeS-anycast_v6_cnt: string, apnic-eyeball_cc_cnt (... 2789 chars omitted)
      child 0, asn: string
      child 1, LACeS-anycast_v4_cnt: string
      child 2, LACeS-anycast_v6_cnt: string
      child 3, apnic-eyeball_cc_cnt: string
      child 4, apnic-eyeball_eyeball_cnt: string
      child 5, apnic-eyeball_gini: string
      child 6, apnic-eyeball_top_cc(frac): string
      child 7, apnic-eyeball_top_cc(num): string
      child 8, apnic-eyeball_top_frac: string
      child 9, caida-asrel_cone_/24_cnt: string
      child 10, caida-asrel_cone_/64_cnt: string
      child 11, caida-asrel_cone_as_cnt: string
      child 12, caida-asrel_cone_as_list: string
      child 13, caida-asrel_customer_cnt: string
      child 14, caida-asrel_custom
...
hars omitted)
      child 0, month: string
      child 1, package: string
      child 2, size: int64
      child 3, sha256: string
      child 4, schema_version: string
      child 5, num_asns: int64
      child 6, source_dates: struct<RIR delegation files: string, APNIC Eyeball: string, OpenIntel Active DNS Measurements: Tranc (... 609 chars omitted)
          child 0, RIR delegation files: string
          child 1, APNIC Eyeball: string
          child 2, OpenIntel Active DNS Measurements: Tranco: string
          child 3, RouteViews Prefix2AS from CAIDA: string
          child 4, IIJ AS Hegemony IPv4: string
          child 5, IIJ AS Hegemony IPv6: string
          child 6, IIJ Traceroute Hegemony: string
          child 7, Internet Intelligence Lab AS2Org: string
          child 8, CAIDA AS Relationship: string
          child 9, MaxMind GeoLite2: timestamp[s]
          child 10, Hypergiants' off-nets estimations: timestamp[s]
          child 11, ISI ANT Censuses of the Internet Address Space: timestamp[s]
          child 12, CAIDA Macroscopic Internet Topology Data Kit: string
          child 13, PeeringDB daily snapshot data from CAIDA: string
          child 14, LACeS Anycast Census: string
          child 15, Censys Universal Internet Bigquery data: string
          child 16, M-Lab NDT Bigquery data: string
          child 17, Merit Network Telescope: string
      child 7, inputs_sha256: string
record_id: null
package_prefix: string
dataset: string
concept_doi: string
to
{'dataset': Value('string'), 'generated_at_utc': Value('string'), 'schema_version': Value('string'), 'concept_doi': Value('string'), 'record_id': Value('null'), 'package_prefix': Value('string'), 'months': List(Value('string')), 'snapshots': List({'month': Value('string'), 'package': Value('string'), 'size': Value('int64'), 'sha256': Value('string'), 'schema_version': Value('string'), 'num_asns': Value('int64'), 'source_dates': {'RIR delegation files': Value('string'), 'APNIC Eyeball': Value('string'), 'OpenIntel Active DNS Measurements: Tranco': Value('string'), 'RouteViews Prefix2AS from CAIDA': Value('string'), 'IIJ AS Hegemony IPv4': Value('string'), 'IIJ AS Hegemony IPv6': Value('string'), 'IIJ Traceroute Hegemony': Value('string'), 'Internet Intelligence Lab AS2Org': Value('string'), 'CAIDA AS Relationship': Value('string'), 'MaxMind GeoLite2': Value('timestamp[s]'), "Hypergiants' off-nets estimations": Value('timestamp[s]'), 'ISI ANT Censuses of the Internet Address Space': Value('timestamp[s]'), 'CAIDA Macroscopic Internet Topology Data Kit': Value('string'), 'PeeringDB daily snapshot data from CAIDA': Value('string'), 'LACeS Anycast Census': Value('string'), 'Censys Universal Internet Bigquery data': Value('string'), 'M-Lab NDT Bigquery data': Value('string'), 'Merit Network Telescope': Value('string')}, 'inputs_sha256': Value('string')})}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2840, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2398, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              snapshot_month: string
              snapshot_id: string
              schema_version: string
              generated_at_utc: string
              artifacts: struct<snapshot_parquet: string, manifest: string, schema_json: string, schema_md: string, data_sour (... 12 chars omitted)
                child 0, snapshot_parquet: string
                child 1, manifest: string
                child 2, schema_json: string
                child 3, schema_md: string
                child 4, data_sources: string
              snapshot: struct<format: string, compression: string, num_asns: int64, columns: list<item: string>, column_sto (... 2902 chars omitted)
                child 0, format: string
                child 1, compression: string
                child 2, num_asns: int64
                child 3, columns: list<item: string>
                    child 0, item: string
                child 4, column_storage_types: struct<asn: string, LACeS-anycast_v4_cnt: string, LACeS-anycast_v6_cnt: string, apnic-eyeball_cc_cnt (... 2789 chars omitted)
                    child 0, asn: string
                    child 1, LACeS-anycast_v4_cnt: string
                    child 2, LACeS-anycast_v6_cnt: string
                    child 3, apnic-eyeball_cc_cnt: string
                    child 4, apnic-eyeball_eyeball_cnt: string
                    child 5, apnic-eyeball_gini: string
                    child 6, apnic-eyeball_top_cc(frac): string
                    child 7, apnic-eyeball_top_cc(num): string
                    child 8, apnic-eyeball_top_frac: string
                    child 9, caida-asrel_cone_/24_cnt: string
                    child 10, caida-asrel_cone_/64_cnt: string
                    child 11, caida-asrel_cone_as_cnt: string
                    child 12, caida-asrel_cone_as_list: string
                    child 13, caida-asrel_customer_cnt: string
                    child 14, caida-asrel_custom
              ...
              hars omitted)
                    child 0, month: string
                    child 1, package: string
                    child 2, size: int64
                    child 3, sha256: string
                    child 4, schema_version: string
                    child 5, num_asns: int64
                    child 6, source_dates: struct<RIR delegation files: string, APNIC Eyeball: string, OpenIntel Active DNS Measurements: Tranc (... 609 chars omitted)
                        child 0, RIR delegation files: string
                        child 1, APNIC Eyeball: string
                        child 2, OpenIntel Active DNS Measurements: Tranco: string
                        child 3, RouteViews Prefix2AS from CAIDA: string
                        child 4, IIJ AS Hegemony IPv4: string
                        child 5, IIJ AS Hegemony IPv6: string
                        child 6, IIJ Traceroute Hegemony: string
                        child 7, Internet Intelligence Lab AS2Org: string
                        child 8, CAIDA AS Relationship: string
                        child 9, MaxMind GeoLite2: timestamp[s]
                        child 10, Hypergiants' off-nets estimations: timestamp[s]
                        child 11, ISI ANT Censuses of the Internet Address Space: timestamp[s]
                        child 12, CAIDA Macroscopic Internet Topology Data Kit: string
                        child 13, PeeringDB daily snapshot data from CAIDA: string
                        child 14, LACeS Anycast Census: string
                        child 15, Censys Universal Internet Bigquery data: string
                        child 16, M-Lab NDT Bigquery data: string
                        child 17, Merit Network Telescope: string
                    child 7, inputs_sha256: string
              record_id: null
              package_prefix: string
              dataset: string
              concept_doi: string
              to
              {'dataset': Value('string'), 'generated_at_utc': Value('string'), 'schema_version': Value('string'), 'concept_doi': Value('string'), 'record_id': Value('null'), 'package_prefix': Value('string'), 'months': List(Value('string')), 'snapshots': List({'month': Value('string'), 'package': Value('string'), 'size': Value('int64'), 'sha256': Value('string'), 'schema_version': Value('string'), 'num_asns': Value('int64'), 'source_dates': {'RIR delegation files': Value('string'), 'APNIC Eyeball': Value('string'), 'OpenIntel Active DNS Measurements: Tranco': Value('string'), 'RouteViews Prefix2AS from CAIDA': Value('string'), 'IIJ AS Hegemony IPv4': Value('string'), 'IIJ AS Hegemony IPv6': Value('string'), 'IIJ Traceroute Hegemony': Value('string'), 'Internet Intelligence Lab AS2Org': Value('string'), 'CAIDA AS Relationship': Value('string'), 'MaxMind GeoLite2': Value('timestamp[s]'), "Hypergiants' off-nets estimations": Value('timestamp[s]'), 'ISI ANT Censuses of the Internet Address Space': Value('timestamp[s]'), 'CAIDA Macroscopic Internet Topology Data Kit': Value('string'), 'PeeringDB daily snapshot data from CAIDA': Value('string'), 'LACeS Anycast Census': Value('string'), 'Censys Universal Internet Bigquery data': Value('string'), 'M-Lab NDT Bigquery data': Value('string'), 'Merit Network Telescope': Value('string')}, 'inputs_sha256': Value('string')})}
              because column names don't match

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

AS feature snapshots (IIL-as-feature-snapshot)

Monthly per-ASN feature tables aggregating signals from routing, DNS, scanning, geolocation, topology, and organization-mapping data sources.

This dataset is the data half of the AS Tagging Toolkit (pip install as-tagging). The methodology is described in the paper Rethinking and Facilitating How We Classify Autonomous Systems by Network Properties (ACM IMC 2026).

How to use

With the toolkit (recommended):

from as_tagging import ASTagging, OnlineSnapshotProvider

provider = OnlineSnapshotProvider()          # reads this dataset
tagger = ASTagging(snapshot_provider=provider, date="2026-07")

Or directly:

  • archives/IIL-as-feature-snapshot.<YYYY-MM>.tar.gz — the full monthly package: one row per ASN (Snappy-compressed parquet) plus manifest.json / schema.json / schema.md / data_sources.txt.
  • snapshots/<YYYY-MM>/ — the same content extracted (parquet as data.parquet), for the datasets library and the dataset viewer.
  • index.json — every released month with size, SHA-256, and per-source input dates.

The dataset is organized per ASN. Each ASN is associated with the features below (feature keys are shown as they appear in the snapshots).

Monthly aggregation strategy

We generate monthly feature snapshots by aggregating data from the sources below.

  • For six measurement-based sources that update daily or at least weekly (RouteViews Prefix-to-AS, IIJ AS Hegemony, APNIC Eyeball, LACeS Anycast Census, OpenIntel Tranco, and IIJ Traceroute Hegemony), we collect all available data within each month and compute the monthly average (mean), to mitigate daily fluctuations and leverage all available measurements.
    Note: For these sources, features with names like *_cnt may be floats because they are monthly averages.

  • For Censys Universal Internet, we use snapshots published on Tuesdays (more comprehensive because they include scans for both hosts and virtual hosts). Due to Google BigQuery quota constraints, we average over two Tuesday snapshots per month.

  • For other sources that update more frequently than monthly, we use simpler strategies:

    • M-Lab NDT: average over speed tests from the first week of each month.
    • PeeringDB and RIR delegation: use a single snapshot per month (no averaging).
    • MaxMind GeoLite2: use the version available on the first day of each month to geolocate prefixes for that month’s snapshot.

Features per data source

  • Delegation (delegation): (single monthly snapshot)

    • delegation_rir (qualitative): The Regional Internet Registry that delegated the AS number.
    • delegation_cc (qualitative): The country code where the AS number was registered.

  • APNIC eyeball dataset (apnic-eyeball): (monthly average over all available days in the month)

    • apnic-eyeball_top_frac: Monthly average of the maximal country eyeball fraction across all countries.
    • apnic-eyeball_eyeball_cnt: Monthly average of total inferred eyeballs across all countries.
    • apnic-eyeball_cc_cnt: Monthly average number of countries in which the AS has eyeballs.
    • apnic-eyeball_gini: Monthly average equality level of country eyeball fraction.
    • apnic-eyeball_top_cc(frac) (qualitative): Country code with maximal eyeball fraction (derived from the aggregated month).
    • apnic-eyeball_top_cc(num) (qualitative): Country code with maximal inferred eyeballs (derived from the aggregated month).

  • Hypergiants' off-nets estimations (hg-offnet): (single monthly snapshot / simple aggregation)

    • hg-offnet_v4addr_cnt: Number of distinct IPv4 addresses hosting hypergiants' off-net servers (for the month).

  • MaxMind GeoLite2 (maxmind-geolite2): (use the version on the first day of the month)

    • maxmind-geolite2_cc_v4_cnt: Number of geolocated countries for originated IPv4 addresses.
    • maxmind-geolite2_cc_v6_cnt: Number of geolocated countries for originated IPv6 addresses.
    • maxmind-geolite2_topfrac_v4: Fraction of IPv4 addresses in the top geolocated country.
    • maxmind-geolite2_topfrac_v6: Fraction of IPv6 addresses in the top geolocated country.
    • maxmind-geolite2_gini_v4: Equality level of IPv4 geolocation distribution.
    • maxmind-geolite2_gini_v6: Equality level of IPv6 geolocation distribution.
    • maxmind-geolite2_cc_v4_dict (qualitative): Country distribution dictionary for IPv4 geolocation.
    • maxmind-geolite2_cc_v6_dict (qualitative): Country distribution dictionary for IPv6 geolocation.
    • maxmind-geolite2_topcc_v4 (qualitative): Top geolocated country code for IPv4.
    • maxmind-geolite2_topcc_v6 (qualitative): Top geolocated country code for IPv6.

  • Censys Universal Internet (censys): (average of two Tuesday snapshots per month)

    • censys_v4addr_cnt: Monthly-averaged number of responsive IPv4 addresses.

    • censys_os_cnt: Monthly-averaged number of distinct scanned operating systems.

    • censys_service_cnt: Monthly-averaged number of distinct scanned services.

    • censys_port_cnt: Monthly-averaged number of distinct scanned ports.

    • censys_voip_cnt: Monthly-averaged number of IPs with open port 5060 or 5061.

    • censys_ics_cnt: Monthly-averaged number of IPs with ports commonly used by industrial control systems.

    • censys_http_cnt: Monthly-averaged number of IPs hosting the HTTP service.

    • censys_ssh_cnt: Monthly-averaged number of IPs with open port 22.

    • censys_auth_cnt: Monthly-averaged number of IPs hosting an authoritative DNS server.

    • censys_forw_cnt: Monthly-averaged number of IPs hosting a forwarding DNS server.

    • censys_recu_cnt: Monthly-averaged number of IPs hosting a recursive DNS server.

    • censys_fdnsname_cnt: Monthly-averaged number of host names detected via forward DNS.

    • censys_rdnsname_cnt: Monthly-averaged number of host names detected via reverse DNS.

    • Top OS (monthly-averaged counts + names):

      • censys_os_1_cnt, censys_os_1_name (qualitative)
      • censys_os_2_cnt, censys_os_2_name (qualitative)
      • censys_os_3_cnt, censys_os_3_name (qualitative)

    • Top Port (monthly-averaged counts + names):

      • censys_port_1_cnt, censys_port_1_name (qualitative)
      • censys_port_2_cnt, censys_port_2_name (qualitative)
      • censys_port_3_cnt, censys_port_3_name (qualitative)

    • Top Service (monthly-averaged counts + names):

      • censys_service_1_cnt, censys_service_1_name (qualitative)
      • censys_service_2_cnt, censys_service_2_name (qualitative)
      • censys_service_3_cnt, censys_service_3_name (qualitative)

  • OpenIntel Active DNS Measurements: Tranco (openintel-tranco): (monthly average over all available data in the month)

    • openintel-tranco_v4addr_cnt: Monthly-averaged number of distinct IPv4 addresses hosting web servers for Tranco domains.
    • openintel-tranco_v6addr_cnt: Monthly-averaged number of distinct IPv6 addresses hosting web servers for Tranco domains.
    • openintel-tranco_topdomain_cnt: Monthly-averaged number of distinct hosted Tranco top domains.

  • CAIDA AS Relationship (caida-asrel): (single monthly snapshot)

    • caida-asrel_provider_cnt: Number of inferred providers.
    • caida-asrel_customer_cnt: Number of inferred customers.
    • caida-asrel_peer_cnt: Number of inferred peers.
    • caida-asrel_cone_/24_cnt: IPv4 space in /24s of the customer cone.
    • caida-asrel_cone_/64_cnt: IPv6 space in /64s of the customer cone.
    • caida-asrel_cone_as_cnt: Number of ASes inferred in the customer cone.
    • caida-asrel_provider_list (qualitative): List of inferred providers.
    • caida-asrel_customer_list (qualitative): List of inferred customers.
    • caida-asrel_peer_list (qualitative): List of inferred peers.
    • caida-asrel_cone_as_list (qualitative): List of ASes inferred in the customer cone.

  • RouteViews Prefix2AS from CAIDA (pfx2as): (monthly average over all available days in the month)

    • pfx2as_/24_cnt: Monthly-averaged IPv4 space in /24s originated by the AS.
    • pfx2as_/64_cnt: Monthly-averaged IPv6 space in /64s originated by the AS.

  • ISI ANT Censuses of the Internet Address Space (isi): (single monthly snapshot / simple aggregation)

    • isi_/24_cnt: Number of /24s with at least one active IP address based on ISI Internet Census.

  • Merit Network Telescope (merit): (single monthly snapshot / simple aggregation)

    • merit_/24_cnt: Maximal number of /24s observed hourly by Merit network telescope within one month.

  • M-Lab NDT (mlab-ndt): (aggregate over the entire month)

    • mlab-ndt_v4addr_cnt: Number of distinct IPv4 client IP addresses that initiated at least one NDT speed test during the month (deduplicated across NDT5 and NDT7).
    • mlab-ndt_v6addr_cnt: Number of distinct IPv6 client IP addresses that initiated at least one NDT speed test during the month (deduplicated across NDT5 and NDT7).
    • mlab-ndt_/24_cnt: Number of distinct IPv4 /24 client subnets (derived from client IPs) that initiated at least one NDT speed test during the month.
    • mlab-ndt_/64_cnt: Number of distinct IPv6 /64 client subnets (derived from client IPs) that initiated at least one NDT speed test during the month.

  • CAIDA Macroscopic Internet Topology Data Kit (ITDK) (caida-itdk): (single monthly snapshot)

    • Aggregate:
      • caida-itdk_router_cnt: Number of routers.
      • caida-itdk_topfrac: Fraction of routers in the top geolocated country.
      • caida-itdk_cc_cnt: Number of geolocated countries for routers.
      • caida-itdk_gini: Equality level of router geolocation.
      • caida-itdk_topcc (qualitative): Top geolocated country code for routers.
    • IPv4-only:
      • caida-itdk_router_cnt_v4
      • caida-itdk_topfrac_v4
      • caida-itdk_cc_cnt_v4
      • caida-itdk_gini_v4
      • caida-itdk_topcc_v4 (qualitative)
    • IPv6-only:
      • caida-itdk_router_cnt_v6
      • caida-itdk_topfrac_v6
      • caida-itdk_cc_cnt_v6
      • caida-itdk_gini_v6
      • caida-itdk_topcc_v6 (qualitative)

  • IIJ AS Hegemony (iij-hege): (monthly average over all available data in the month)

    • iij-hege_global_hege_v4: Monthly average AS hegemony based on global IPv4 BGP data.
    • iij-hege_global_hege_v6: Monthly average AS hegemony based on global IPv6 BGP data.

  • IIJ Traceroute Hegemony (iij-tr-hege): (monthly average over all available data in the month)

    • iij-tr-hege_ix_cnt: Monthly-averaged number of peered IXPs (traceroute-based).
    • iij-tr-hege_peers_cnt: Monthly-averaged number of peered ASNs through IXPs (traceroute-based).
    • iij-tr-hege_ix_hegemony: Monthly-averaged summation of IXP-peered ASNs hegemony.

  • Internet Intelligence Lab AS2Org (inetintel-as2org): (single monthly snapshot)

    • inetintel-as2org_sibling_cnt: Number of inferred sibling ASes.
    • inetintel-as2org_sibling_list (qualitative): List of inferred sibling ASes.

  • PeeringDB (pdb): (single monthly snapshot)

    • pdb_ix_cnt: Number of peered IXPs based on PeeringDB.
    • pdb_website (qualitative): Website URL based on PeeringDB.

  • LACeS Anycast Census (LACeS-anycast): (monthly average over all available data in the month)

    • LACeS-anycast_v4_cnt: Monthly-averaged number of distinct IPv4 addresses detected as anycast.
    • LACeS-anycast_v6_cnt: Monthly-averaged number of distinct IPv6 addresses detected as anycast.

License

The AS feature-snapshot data is distributed under Georgia Tech's Acceptable Use Agreement. See LICENSE in the GitHub repository for the full terms.
Downloads last month
73
Total file size:
1.19 GB