Module temporal_join

Temporal join — find time windows where entities are semantically close.

A temporal join finds periods where two (or more) entities converge in embedding space. Unlike point-level kNN, this captures sustained proximity.

Example

Two users who sporadically post similar content won’t trigger a join, but users who consistently converge over a week-long window will.

Structs

GroupJoinResult

A time window during which a subset of a group converged.

TemporalJoinResult

A time window during which two entities were semantically close.

Functions

find_largest_epsilon_cluster 🔒

Find the largest cluster of entities where all pairs are within epsilon.

group_temporal_join

Find time windows where at least min_entities from the group converge.

mean_pairwise_distance 🔒

Mean pairwise distance among a subset of entities.

merge_group_results 🔒

Merge consecutive group join results that overlap in time.

temporal_join

Find time windows where entities A and B are within distance epsilon.

window_distances 🔒

Compute min and mean distances between A’s and B’s points in a time window.