Typical Workflow

1. Choose inputs

Load event-table records and identify the study context (conditions, actors, outcomes).

If you are working from design-research-experiments exports, start with events.csv and use Experiments-To-Analysis Handoff for the canonical validation and rejoin flow back to runs.csv and evaluations.csv.

2. Instantiate core objects

Create unified-table validation/derivation config and choose analysis-family entry points.

3. Execute or inspect

Run sequence, language, embedding-map, and/or statistical workflows.

4. Capture artifacts

Write structured summaries, optional exports, and run-manifest metadata.

5. Connect to the next library

Feed interpreted findings back to design-research-experiments for protocol refinement and to agent/problem selection decisions for future studies.

Choosing Analysis Families

Sequence analysis is best when order and transitions matter. Language analysis is best when semantic change or discourse properties matter. Embedding maps are best when embedding geometry, trajectories, and clustering structure matter. Statistical workflows are best when inferential claims and effect-size estimation matter.