Document Clustering

Cluster documents by text content (or bibliographic coupling).

Document Clustering

The Document Clustering widget.

Overview

Groups documents into clusters from their text (TF-IDF) or, optionally, from bibliographic coupling (shared references), adding a Cluster column. A hard-assignment complement to Topic Modeling (soft topic weights).

Inputs

  • Data (Table) — bibliographic data.

Outputs

  • Data (Table) — input with a Cluster column.

  • Cluster Summary (Table) — cluster sizes.

  • Selected Documents (Table) — documents in the selected cluster(s).

Controls

  • Cluster by (text) — the text field used for clustering.

  • Method — clustering algorithm (k-means, agglomerative…).

  • N clusters (0 = auto) — number of clusters, or 0 to choose automatically.

  • Bibliographic coupling (optional) — cluster on shared references instead of/with text.

Actions: Cluster.