Book — Research and evaluation notes

These are Advanced design and evaluation notes for readers with basic Python and database skills, not a course in research methods or a required setup path. Start with Your first research project for an integrated social-science example, or the basic graph for graph construction and queries.

The notes distinguish supported API behavior from methodological choices and workload-specific measurements that need their own evidence. A runnable graph query does not establish valid interpretation, model quality, or usability.

Note	Worked guide
Knowledge graph construction	Source-linked mentions
Network analysis	Notebook network analysis
Analyst verbs at scale	Native graph algorithms
LLM-powered workflows	Extraction and human review
LLM context building	Source-linked extraction
Agent grounding	Tool and capability graphs
Search and entity resolution	Candidate lookup
Genealogy modelling	Source-qualified family relationships

Current signatures live in the API reference. Measurement results belong on their owning issue; these notes describe methods and limitations rather than offering unqualified latency or quality guarantees.

Research history from the terminal

Versions are commits, Branches are refs, and a tag names a Version. The terminal workflow (gf branch, gf commit, gf tag, gf push, gf pull, gf log, gf status, gf switch) and the one meaning of --ref are described in Share with gf; the decision is ADR 0061.