Clusteringconcept
Grouping items so similar ones share a cluster — in knowledge compilation, grouping embeddings by similarity into topic clusters.
Clustering turns a flat embedding set into navigable topic groups (the cluster graph view), a cheap runtime structure.
Where it appears
- Knowledge Compiler clustering phase (Phase 5)
Descend
Prerequisites
Dependencies
- —
Foundations
Historical evolution
Connected-components clustering on a similarity graph (threshold >= 0.1) with centroid averaging.
Implementation details
SKCE's concept graph is edge-weighted by co-occurrence rather than geometric clustering, but the principle is shared.
Sources
- [2026-07-11-knowledge-compiler-compiling-human-knowledge-into-static-semantic-artifacts] 2026-07-11-knowledge-compiler-compiling-human-knowledge-into-static-semantic-artifacts
- [2026-07-11-knowledge-compiler-compiling-human-knowledge-into-static-semantic-artifacts] 2026-07-11-knowledge-compiler-compiling-human-knowledge-into-static-semantic-artifacts