Concept Taxonomy¶
The 379 concepts in the AI for EE Research learning graph are organized into
17 categories. Each concept has exactly one primary category, identified by a
3-6 letter TaxonomyID that is used in learning-graph.csv and as the group
key in learning-graph.json. The categories follow the course's teaching
order: foundations first, then how models work, then the Markdown and skill
building blocks, then agents and orchestration, then the five research tasks,
and finally verification, ethics, and the capstone. The always-on personal
agents and personal knowledge graph categories were added afterward and are
numbered last so that existing concept IDs did not change.
No category holds more than 10% of the concepts, well under the 30% ceiling.
Categories¶
FOUND — Foundations of AI-Assisted Research¶
Concepts 1-20. The research workflow, the five target tasks, the core vocabulary of generative AI and agents, and the central idea of the researcher as orchestrator.
LLM — How Language Models Work¶
Concepts 21-50. Tokens, context windows, prompts and completions, training data, hallucination, model selection and cost, retrieval, and the interfaces researchers use to reach a model.
PROMPT — Prompting and Context Engineering¶
Concepts 51-80. Writing clear instructions, output formats and acceptance criteria, choosing and compressing context, grounding, prompt libraries, testing, and multi-step prompting.
MDKB — Markdown and Project Knowledge¶
Concepts 81-110. Markdown syntax, project instruction files such as
AGENTS.md and CLAUDE.md, notes, specifications, logs, knowledge bases, and
the Git practices that keep them reliable.
SKILL — Agent Skills¶
Concepts 111-135. Designing, writing, scoping, testing, versioning, sharing, and maintaining reusable agent skills for repeatable research tasks.
AGENT — Agents and Tool Use¶
Concepts 136-170. The agent loop, tool calls, file access, shell commands, code execution, permissions, subagents, hooks, external data, autonomy, and agent failure modes.
ORCH — Agent Teams and Orchestration¶
Concepts 171-205. Planning and execution agents, roles, handoffs, task queues, review checkpoints, delegation, and the 20-planning-agent, 10-execution-agent orchestration model.
LIT — Literature Search and Synthesis¶
Concepts 206-225. Search strategy, paper triage and summarization, citation chaining, reference management, citation verification, and evidence synthesis.
DATA — Data Analysis and Scientific Coding¶
Concepts 226-255. Data cleaning, analysis scripts, plotting, testing and debugging, simulation and signal-processing code, HDL, and reproducible environments.
BRAIN — Brainstorming and Problem Framing¶
Concepts 256-265. Framing problems, generating and ranking ideas, developing hypotheses, designing experiments, and checking novelty.
MANU — Manuscript Preparation¶
Concepts 266-284. Paper structure, drafting and revision, author voice, LaTeX, figures, references, arXiv submission, peer review response, and authorship.
COMM — MicroSims and Research Communication¶
Concepts 285-296. Slide decks, MicroSims, interactive simulations, intelligent textbooks, learning graphs, and teaching from research.
VERIF — Verification and Reproducibility¶
Concepts 297-312. Verification and validation, independent and cross-model checks, result replication, sycophancy and overconfidence, audit trails, and error tracing.
ETHIC — Responsible and Ethical Use¶
Concepts 313-336. Academic integrity, citation and disclosure of AI use, privacy, intellectual property, bias, institutional and journal policy, accountability, and over-reliance.
CASE — Case Studies and Capstone¶
Concepts 337-350. The MicroSims arXiv paper, the intelligent textbook project, lessons learned, agent team design, and the capstone project.
PERS — Always-On Personal Agents¶
Concepts 351-369. Always-on, cloud-hosted and self-hosted personal agents: persistent cloud computers, delegated sign-in, taught routines, the individual products (Grok Bot, Dots, Muse, Claude Cowork, OpenClaw, and Hermes Agent), blast radius, usage metering, comparison criteria, vendor-claim verification, research-data exposure, overnight audits, and agent selection. See the appendix Personal Agent Software.
PKG — Personal Knowledge Graphs¶
Concepts 370-379. Knowledge graphs, entities and relationships, personal knowledge graphs built from a researcher's own notes and papers, graph schema, backlinks, entity extraction, graph querying, graph retrieval augmented generation, knowledge graph memory for agents, and the privacy of personal knowledge.
TaxonomyID Reference¶
| TaxonomyID | Category Name | Concept ID Range |
|---|---|---|
| FOUND | Foundations of AI-Assisted Research | 1-20 |
| LLM | How Language Models Work | 21-50 |
| PROMPT | Prompting and Context Engineering | 51-80 |
| MDKB | Markdown and Project Knowledge | 81-110 |
| SKILL | Agent Skills | 111-135 |
| AGENT | Agents and Tool Use | 136-170 |
| ORCH | Agent Teams and Orchestration | 171-205 |
| LIT | Literature Search and Synthesis | 206-225 |
| DATA | Data Analysis and Scientific Coding | 226-255 |
| BRAIN | Brainstorming and Problem Framing | 256-265 |
| MANU | Manuscript Preparation | 266-284 |
| COMM | MicroSims and Research Communication | 285-296 |
| VERIF | Verification and Reproducibility | 297-312 |
| ETHIC | Responsible and Ethical Use | 313-336 |
| CASE | Case Studies and Capstone | 337-350 |
| PERS | Always-On Personal Agents | 351-369 |
| PKG | Personal Knowledge Graphs | 370-379 |