RAGMap question examples
Date: 2026-06-01
Purpose: Natural-language prompts a future RAGMap interface should support, with graph + retrieval requirements and display safety.
Legend — public display:
| Code |
Meaning |
| Public |
Answer may use MkDocs pages + include_in_public_site data only |
| Internal |
May use full graph (include_in_graph), reports, link YAML evidence |
| Restricted |
May reference local grant PDFs, ORS purchasing, or unpublished proposals — permission required |
Discovery and routing
Q1. Who should I talk to about environmental sensing at DePaul?
| Field |
Detail |
| Intended answer type |
Ranked people list with roles and facility names |
| Graph query |
People with CONNECTED_TO_RESOURCE to CSH/CDM resources where equipment taxonomy or resource name matches sensing keywords; prefer evidence_level=direct |
| Evidence retrieval |
Resource Markdown (CSH instrumentation), evidence_text on edges, faculty profiles |
| Public display |
Internal (graph edges may include non-public contacts) |
Q2. What resources support student-built electronics projects?
| Field |
Detail |
| Intended answer type |
Resource list + access notes + related courses |
| Graph query |
Resources with equipment categories; optional COURSE_USES_RESOURCE (phase 2) |
| Evidence retrieval |
resource_equipment.yaml, curriculum pages, access fields |
| Public display |
Public if access URLs cited; internal for weak edges |
Q3. Which facilities could support a proposal involving imaging and machine learning?
| Field |
Detail |
| Intended answer type |
Facility shortlist + suggested grant/publication precedents |
| Graph query |
Resource neighborhood: grants SUPPORTS_RESOURCE, pubs DOCUMENTS_RESOURCE, outputs OUTPUT_USES_RESOURCE |
| Evidence retrieval |
Grant titles, HPCC/VIDA/IRL pages, publication relevance text |
| Public display |
Internal for proposal drafting |
Evidence and documentation
Q4. What evidence do we have that Cinespace is used for student or faculty outputs?
| Field |
Detail |
| Intended answer type |
Cited bullet list (OUT-*, people, URLs) |
| Graph query |
MATCH (o:Output)-[:OUTPUT_USES_RESOURCE]->(r:Resource {resource_id:'CDM-016'}) + person links |
| Evidence retrieval |
Output descriptions, SCA/Project Bluelight pages, outputs-quality-review.md |
| Public display |
Public when citing public URLs only |
Q5. Which grants mention facilities or equipment?
| Field |
Detail |
| Intended answer type |
Grant table with linked resources |
| Graph query |
Grant–SUPPORTS_RESOURCE–Resource where include_in_graph=true |
| Evidence retrieval |
grants.yaml narratives, evidence_file on grant–resource links |
| Public display |
Internal (some grants from internal CSV) |
Q6. Why does resource CDM-002 appear in the refresh report with warnings?
| Field |
Detail |
| Intended answer type |
Explanation + recommended next documentation step |
| Graph query |
Count person edges where include_in_graph=true (= 0) |
| Evidence retrieval |
full-refresh-report.md warning disposition, CDM-002 Markdown |
| Public display |
Internal |
Gaps and enrichment
Q7. Which resources are likely underdocumented?
| Field |
Detail |
| Intended answer type |
Gap list (equipment but no people, no grant link, stale verification) |
| Graph query |
Anti-join patterns: equipment index ∩ zero direct person edges; no grant–resource |
| Evidence retrieval |
Refresh warnings, verification review, internal-data-wishlist.md |
| Public display |
Internal |
Q8. What internal ORS data would improve this answer?
| Field |
Detail |
| Intended answer type |
Field-level wishlist tied to specific resource/grant IDs |
| Graph query |
Triggered by missing properties on Grant or Resource nodes in subgraph |
| Evidence retrieval |
internal-data-wishlist.md, grants layer review |
| Public display |
Restricted |
Q9. Which people connect sustainability, design, and sensing?
| Field |
Detail |
| Intended answer type |
Bridge people with topic/resource evidence per theme |
| Graph query |
Multi-prefix person–resource paths; optional topic nodes (phase 2) |
| Evidence retrieval |
Profiles, publication relevance text — not keyword guess alone |
| Public display |
Internal — high false-inference risk |
Equipment and operations
Q10. What equipment appears to be associated with chemistry instrumentation?
| Field |
Detail |
| Intended answer type |
Equipment lines grouped by resource (CSH-003, CSH-008, etc.) |
| Graph query |
Resource nodes with prefix CSH + equipment join (phase 2) |
| Evidence retrieval |
resource_equipment.yaml, CSH facility pages |
| Public display |
Public |
Q11. Who operates CSH-008 if the graph has no named person?
| Field |
Detail |
| Intended answer type |
Honest “generic contact only” + link to chemistry IT |
| Graph query |
Confirm zero include_in_graph person edges |
| Evidence retrieval |
Resource page, refresh warning rationale |
| Public display |
Public |
Collaboration and hubs
Q12. Who bridges CDM and CSH technical resources with strong evidence?
| Field |
Detail |
| Intended answer type |
Person shortlist with edge types and resource IDs |
| Graph query |
Cross-prefix count; filter evidence_level=direct |
| Evidence retrieval |
Per-edge evidence_text |
| Public display |
Internal |
Q13. Is UNI-001 the main hub for my topic?
| Field |
Detail |
| Intended answer type |
No — explain ORS anchoring; redirect to specific labs |
| Graph query |
False-hub audit: exclude UNI-001 from centrality |
| Evidence retrieval |
person_link_rules.py policy, graph-readiness review |
| Public display |
Internal |
Q14. Which collaborators are safe to name in a federal proposal?
| Field |
Detail |
| Intended answer type |
Only direct / documented collaborates with citations |
| Graph query |
Exclude affiliation-only; require publication or grant co-path |
| Evidence retrieval |
Evidence files, DOIs, grant roles |
| Public display |
Restricted if using unpublished grant text |
Student and curriculum
Q15. Which courses document use of a specific makerspace?
| Field |
Detail |
| Intended answer type |
Course IDs + documented link URLs |
| Graph query |
COURSE_USES_RESOURCE (phase 2) |
| Evidence retrieval |
course_resource_links.yaml, CUR pages |
| Public display |
Public |
Q16. What outputs demonstrate IRL or game lab activity?
| Field |
Detail |
| Intended answer type |
Output catalog with people and resources |
| Graph query |
Outputs linked to CDM-001/CDM-003/etc. |
| Evidence retrieval |
outputs.yaml, news URLs |
| Public display |
Public |
Q17. How many graph-included edges exist per layer?
| Field |
Detail |
| Intended answer type |
Count table matching full refresh |
| Graph query |
Aggregate relationship counts with include_in_graph=true |
| Evidence retrieval |
full-refresh-report.md |
| Public display |
Public |
Q18. What changed since the last refresh?
| Field |
Detail |
| Intended answer type |
Diff summary (if git or report history available) |
| Graph query |
Optional — compare CSV snapshots |
| Evidence retrieval |
Git log, prior full-refresh-report |
| Public display |
Internal |
Implementation checklist (per question)
- Parse intent → select use case from graph-analysis-use-cases.md.
- Run parameterized Cypher (or pandas equivalent) with default weak-edge filters.
- Collect
evidence_file / resource paths for each result row.
- Retrieve and chunk text; attach to prompt with node IDs.
- Generate answer with mandatory citations; label confidence and gaps.
- Enforce public display policy before showing in any future public UI.