Embedding Landscape

2-D semantic map of documents from text embeddings.

Embedding Landscape

The Embedding Landscape widget.

Overview

Embeds documents from their text, projects them to 2-D, and clusters them into a semantic landscape where nearby points are topically similar. A content-based complement to keyword co-occurrence maps. Outputs the 2-D coordinates and a cluster label per document.

Inputs

  • Data (Table) — data with a text column.

Outputs

  • Coordinates (Table) — 2-D coordinates + cluster per document.

Controls

  • Text column — the field embedded (e.g. Abstract).

  • Clusters — number of clusters to form.

Actions: Build Landscape.