HITS

UDTF: cugraph_hits

Official cuGraph reference: C API

Compute mutually reinforcing hub and authority scores: strong hubs point to strong authorities, and strong authorities are linked from strong hubs.

Quickstart

The call below supplies edges from registered relation target_edges with canonical src and dst columns and may include weight. Substitute your own registered relations.

SELECT *
FROM cugraph_hits(
  edges => (SELECT src, dst FROM target_edges)
);

Inputs

Every relation is a named parenthesized SELECT subquery. The required edges role uses canonical src and dst columns; every role, its canonical columns, and their accepted Arrow types are listed under Relation arguments. Metadata validation resolves registered tables named in its JSON request.

Endpoint columns accept numeric Int32, Int64 vertex IDs or logical string Utf8, LargeUtf8, Utf8View vertex IDs; string vertex-identity outputs are canonicalized to Utf8 (native mapping Int64) while scores, distances, counts, coordinates, and opaque labels stay numeric. The shared vertex-ID contract is summarized in Vertex ID support; the concrete call-specific schema comes from gpu_validate_call.

Logical string side-input limitations:

• edge ID columns and edge-ID predicate side inputs are not supported for logical string graphs

Arguments and options

Relation arguments

Argument	Required	Columns	Description
edges	yes	src, dst: Int32, Int64, Utf8, LargeUtf8, Utf8View weight (optional): Float32, Float64 edge_id (optional): Int32, Int64	edge relation with canonical src and dst columns, plus optional weight and edge_id columns

Every edge_id column must have the integer type of src and dst; string-keyed graphs accept no edge IDs.

Named value arguments

Option	Type	Default	Constraints	Description
epsilon	number	0.00001	> 0	positive convergence tolerance
max_iterations	integer	100	min 1; max 4294967295	maximum iteration count, at least 1
normalize	boolean	false		whether HITS scores are normalized

Graph construction options

Graph construction follows the shared defaults (directed=true, renumbering, python_cugraph policy) documented in Graph Construction Options.

Output

Column	Type	Nullable	Description
vertex	Int64|Utf8	no	Vertex receiving HITS scores.
hub_score	Float64	no	HITS hub score for the vertex.
authority_score	Float64	no	HITS authority score for the vertex.

These are generic descriptor schemas; run gpu_validate_call to get the concrete, table-specific output schema.

Examples

This example runs on the citation network demo dataset.

Which papers act as reading guides, and which works are their shared foundations?

HITS scores two citation roles through mutually reinforcing links. A paper's hub score reflects the authority of works it cites; its authority score reflects how strongly it is cited by hubs. This query ranks the top four vertices for each role and joins both lists back to paper titles and years.

CREATE OR REPLACE TABLE hits_snapshot AS
SELECT vertex, hub_score, authority_score
FROM cugraph_hits(edges => (SELECT src, dst FROM citation_edges));

WITH guides AS (
  SELECT vertex, hub_score AS score, 'Reading guide' AS role
  FROM hits_snapshot
  ORDER BY hub_score DESC, vertex
  LIMIT 4),
foundations AS (
  SELECT vertex, authority_score AS score, 'Foundation' AS role
  FROM hits_snapshot
  ORDER BY authority_score DESC, vertex
  LIMIT 4)
SELECT h.role, p.year, p.title
FROM (
  SELECT * FROM guides
  UNION ALL
  SELECT * FROM foundations
) h
JOIN papers p ON p.paper_id = h.vertex
ORDER BY h.role, h.score DESC, h.vertex;

role	year	title
Foundation	2004	Distinctive Image Features from Scale-Invariant Keypoints
Foundation	2014	VERY DEEP CONVOLUTIONAL NETWORKS FOR LARGE-SCALE IMAGE RECOGNITION
Foundation	2012	ImageNet Classification with Deep Convolutional Neural Networks
Foundation	2005	Histograms of oriented gradients for human detection
Reading guide	2019	Deep Learning for Generic Object Detection: A Survey
Reading guide	2018	Deep Learning for Generic Object Detection: A Survey.
Reading guide	2015	Recent Advances in Convolutional Neural Networks
Reading guide	2019	Object Detection With Deep Learning: A Review

One snapshot supplies both rankings. The returned rows identify each paper's role, year, and title. These scores describe citation structure; they do not prove research quality.

Limits

No algorithm-specific limitations.

To dry-run validate relation metadata, column types, and options without execution, see gpu_validate_call.