Education Topic Decomposition
Goal
Turn a broad education topic into 3-8 researchable directions with evidence-backed subtopics, keywords, possible populations, methods, theories, and gaps.
Inputs
Ask for missing essentials only if absent:
- Broad topic or problem, e.g. "AI feedback in middle-school writing"
- Education level/context, e.g. K-12, higher education, vocational education
- Target paper type, e.g. empirical article, thesis, review, intervention study
- Region/language preference, if relevant
- Constraints, e.g. available data, time, sample access
Core Workflow
- Restate the topic as a research landscape.
- Generate 5-8 perspectives using STORM-style questioning:
- Learner outcome perspective
- Teacher practice perspective
- Technology/tool perspective
- Equity/inclusion perspective
- Assessment perspective
- Policy/institution perspective
- Methodological perspective
- For each perspective, generate search keywords in English and Chinese if useful.
- Search OpenAlex or Semantic Scholar for recent and highly cited work.
- Identify clusters: mature topics, emerging topics, underexplored intersections, feasible empirical directions.
- Output a topic-decomposition table and recommend 2-3 best directions.
Tool Calls
OpenAlex Works Search
Use when no API key is available or broad topic exploration is enough.
curl "https://api.openalex.org/works?search=AI%20feedback%20writing%20education&filter=from_publication_date:2021-01-01&per-page=10&mailto=YOUR_EMAIL"
Useful fields:
display_namepublication_yearcited_by_countconceptstopicsabstract_inverted_indexreferenced_works
Semantic Scholar Search
Use when you need abstracts, citation counts, influential citations, and recommendations.
curl "https://api.semanticscholar.org/graph/v1/paper/search?query=AI%20feedback%20writing%20education&limit=10&fields=title,year,abstract,citationCount,referenceCount,authors,fieldsOfStudy"
If an API key exists:
curl -H "x-api-key: $S2_API_KEY" "https://api.semanticscholar.org/graph/v1/paper/search?query=AI%20feedback%20writing%20education&limit=10&fields=title,year,abstract,citationCount,authors"
STORM
Use when the user wants a deep topic outline or multi-perspective prewriting.
Install:
git clone https://github.com/stanford-oval/storm.git
cd storm
pip install -r requirements.txt
Use STORM as an inspiration engine: generate perspectives, questions, and a structured outline, then verify claims with scholarly search.
Output Format
Return:
- Topic landscape summary
- Decomposition table:
| Direction |
Core Problem |
Key Concepts |
Search Terms |
Possible Research Questions |
Feasible Data |
Suitable Methods |
Evidence Signals |
Risk |
- Top 2-3 recommended directions with reasons:
- novelty
- feasibility
- available literature
- fit with education research norms
Quality Rules
- Do not present a direction as novel unless search evidence supports that it is underexplored.
- Separate "interesting topic" from "researchable topic".
- Prefer directions where variables, population, method, and data source can be named.
- For education papers, always map the topic to learner, teacher, curriculum, assessment, institution, or technology context.
1---2name: education-topic-decomposition3description: Use when decomposing a broad education research topic into focused subtopics, searchable concepts, research gaps, candidate study designs, and writing directions. Best for thesis/article topic selection, narrowing a vague idea, or turning an education policy/practice issue into researchable tracks using STORM-style multi-perspective questioning plus OpenAlex/Semantic Scholar evidence search.4---56# Education Topic Decomposition78## Goal910Turn a broad education topic into 3-8 researchable directions with evidence-backed subtopics, keywords, possible populations, methods, theories, and gaps.1112## Inputs1314Ask for missing essentials only if absent:1516- Broad topic or problem, e.g. "AI feedback in middle-school writing"17- Education level/context, e.g. K-12, higher education, vocational education18- Target paper type, e.g. empirical article, thesis, review, intervention study19- Region/language preference, if relevant20- Constraints, e.g. available data, time, sample access2122## Core Workflow23241. Restate the topic as a research landscape.252. Generate 5-8 perspectives using STORM-style questioning:26 - Learner outcome perspective27 - Teacher practice perspective28 - Technology/tool perspective29 - Equity/inclusion perspective30 - Assessment perspective31 - Policy/institution perspective32 - Methodological perspective333. For each perspective, generate search keywords in English and Chinese if useful.344. Search OpenAlex or Semantic Scholar for recent and highly cited work.355. Identify clusters: mature topics, emerging topics, underexplored intersections, feasible empirical directions.366. Output a topic-decomposition table and recommend 2-3 best directions.3738## Tool Calls3940### OpenAlex Works Search4142Use when no API key is available or broad topic exploration is enough.4344```bash45curl "https://api.openalex.org/works?search=AI%20feedback%20writing%20education&filter=from_publication_date:2021-01-01&per-page=10&mailto=YOUR_EMAIL"46```4748Useful fields:4950- `display_name`51- `publication_year`52- `cited_by_count`53- `concepts`54- `topics`55- `abstract_inverted_index`56- `referenced_works`5758### Semantic Scholar Search5960Use when you need abstracts, citation counts, influential citations, and recommendations.6162```bash63curl "https://api.semanticscholar.org/graph/v1/paper/search?query=AI%20feedback%20writing%20education&limit=10&fields=title,year,abstract,citationCount,referenceCount,authors,fieldsOfStudy"64```6566If an API key exists:6768```bash69curl -H "x-api-key: $S2_API_KEY" "https://api.semanticscholar.org/graph/v1/paper/search?query=AI%20feedback%20writing%20education&limit=10&fields=title,year,abstract,citationCount,authors"70```7172### STORM7374Use when the user wants a deep topic outline or multi-perspective prewriting.7576Install:7778```bash79git clone https://github.com/stanford-oval/storm.git80cd storm81pip install -r requirements.txt82```8384Use STORM as an inspiration engine: generate perspectives, questions, and a structured outline, then verify claims with scholarly search.8586## Output Format8788Return:89901. Topic landscape summary912. Decomposition table:9293| Direction | Core Problem | Key Concepts | Search Terms | Possible Research Questions | Feasible Data | Suitable Methods | Evidence Signals | Risk |94|---|---|---|---|---|---|---|---|---|95963. Top 2-3 recommended directions with reasons:97 - novelty98 - feasibility99 - available literature100 - fit with education research norms101102## Quality Rules103104- Do not present a direction as novel unless search evidence supports that it is underexplored.105- Separate "interesting topic" from "researchable topic".106- Prefer directions where variables, population, method, and data source can be named.107- For education papers, always map the topic to learner, teacher, curriculum, assessment, institution, or technology context.
ilog3