Proposal Writing Workflow Using the ECC/SHCC/EGC Atlas

Status

• Workflow layer: 09_ai_workflows.
• Purpose: use the ECC Research Atlas to draft source-grounded research proposal backgrounds, especially NRF-style proposals.
• Related proposal direction: Physical AI is premature as a stand-alone claim; the near-term direction is a Wiki-RAG based materials knowledge system plus persistent experimental data infrastructure for AI-assisted mixture design.
• Critical rule: proposal claims must be grounded in Atlas source paths, claim-evidence rows, or lab positioning notes.

Core proposal logic

Global ECC/EGC knowledge gap
  -> source-grounded evidence from Atlas
    -> Lee lab positioning and readiness
      -> research need
        -> proposed methodology
          -> expected scientific and societal impact

1. Proposal topic definition

Purpose

Define the proposal topic narrowly enough that the Atlas can retrieve the correct evidence.

Required inputs

proposal_topic:
target_material_system:
main_mechanism:
performance_goal:
sustainability_or_application_goal:
Lee_lab_axis:
proposal_status: idea | draft | NRF | internal planning

Recommended topic categories

Topic definition template

This proposal aims to develop <material/system> by controlling <mechanism/design variable> to achieve <performance goal>, while addressing <sustainability/application need>. The work is positioned within <global Atlas lineage> and builds on <Lee lab research axis>.

NRF AI/Wiki-RAG example

This proposal aims to build a Wiki-RAG based materials knowledge system for ECC/EGC mixture design by linking source-grounded literature evidence, lab publication data, material variables, and performance metrics. The goal is to generate experimentally testable mixture hypotheses while building the persistent data infrastructure required before Physical AI can be meaningfully pursued.

2. Global literature gap extraction

Retrieval targets

Start with global nodes and extension matrices:

index.md
02_concepts/strain_hardening_criteria.md
02_concepts/fiber_bridging_law.md
02_concepts/flaw_design.md
02_concepts/extreme_ductility_ecc.md
04_material_systems/green_ecc.md
04_material_systems/geopolymer_ecc.md
04_material_systems/self_healing_ecc.md
07_visualization/extreme_ductility_extension_claim_evidence_matrix.csv
07_visualization/sustainable_ecc_extension_claim_evidence_matrix.csv
07_visualization/foundational_papers_claim_evidence_matrix.csv

Gap extraction questions

1. What has the global literature already established?
2. What remains difficult or unresolved?
3. Is the gap mechanistic, material-system, durability, sustainability, cost, or implementation-related?
4. Which evidence rows show the state of the art?
5. Which evidence rows show limitations or caution?

Gap statement template

Although prior ECC/SHCC research has established <known foundation>, existing studies still face <specific limitation>. In particular, <material/system> requires <missing mechanism/data/workflow> to achieve <target performance> under <practical constraint>. This creates a need for <proposed research direction>.

Example: low-fiber EGC / EPS flaw design

Although ECC micromechanics has established that distributed cracking requires adequate fiber bridging and matrix crack activation, reducing PE fiber content remains difficult because the PSH margin becomes smaller. A key gap is how to deliberately tailor the flaw population and matrix cracking threshold so that low-fiber EGC can retain strain-hardening without relying on high fiber dosage.

Example: Wiki-RAG materials AI

Although a large body of ECC/EGC literature reports binder, fiber, flaw, curing, tensile, durability, and sustainability data, these data remain scattered across papers and are not readily usable for AI-assisted material design. Physical AI is premature without a persistent source-grounded data infrastructure. A Wiki-RAG Atlas can fill this gap by linking claims, evidence, material variables, and performance metrics in a traceable system.

3. Lee lab positioning extraction

Retrieval priority

For proposals connected to Professor Lee's research, retrieve in this order:

06_lab_position/our_lab_position_map.md
06_lab_position/<axis>_position.md
07_visualization/lab_to_global_lineage_map.md
06_lab_position/by_lee_publication_mapping.csv
07_visualization/by_lee_lab_publications_claim_evidence_matrix.csv
03_papers/by_lee_lab_publications/

Lab positioning fields to extract

Lee_lab_axis
representative_papers
key_metrics
mechanism_contribution
material_system_contribution
strategic_positioning_sentence
cautions_or_overclaim_limits

Positioning statement template

Professor Lee's research group is positioned in this proposal as <specific role>, based on prior work on <representative research axis>. The lab's contribution is not merely <generic description>, but <specific mechanism/material/system contribution> supported by <source-grounded evidence>.

Example: cementless EGC

Professor Lee's group is positioned as a sustained contributor to transferring ECC micromechanics into cementless AAS and geopolymer binder systems. The lab's contribution is not binder substitution alone, but the adaptation of matrix chemistry, fiber/interface behavior, and PSH design to retain strain-hardening and crack-width control.

Example: recycled selvage fiber ECC

Professor Lee's group is positioned as developing a circular-economy ECC route in which high-performance PE textile waste is processed into structural crack-bridging fibers. This contribution links recycled fiber use to verified tensile ductility, crack-width control, and cost reduction rather than treating waste fibers as low-grade fillers.

4. Claim-evidence support table

Every proposal background should include an internal support table before writing prose.

Table template

| Proposal claim | Evidence source | Evidence summary | Verification level | Use in proposal |
|---|---|---|---|---|
| <claim> | `<path>` | <short evidence> | verified_from_pdf / source_note_pending_pdf | background / novelty / method / impact |

Required evidence sources

Use at least one from each category when possible:

1. Global anchor or foundational mechanism: - 00_sources/victor_li_book/ - 07_visualization/foundational_papers_claim_evidence_matrix.csv
2. Topic-specific extension: - 07_visualization/extreme_ductility_extension_claim_evidence_matrix.csv - 07_visualization/sustainable_ecc_extension_claim_evidence_matrix.csv
3. Lee lab positioning: - 06_lab_position/*.md - 07_visualization/by_lee_lab_publications_claim_evidence_matrix.csv

Example support table row

| Low-fiber EGC should be framed as flaw/matrix design rather than simply fiber reduction. | `06_lab_position/low_fiber_egc_position.md`; `02_concepts/flaw_design.md` | Intentional flaws can lower matrix crack activation threshold if fiber bridging remains sufficient. | source-grounded positioning / foundational support | novelty + methodology |

5. Research need paragraph template

Template

Infrastructure materials are required to simultaneously reduce environmental burden, improve durability, and maintain mechanical reliability. ECC/SHCC provides a micromechanical framework for tensile strain-hardening and crack-width control, but <specific problem> remains unresolved. Existing studies have shown <established evidence>, yet <gap> limits practical deployment. Therefore, there is a clear need to develop <proposed system/workflow> that can <target function> while preserving <ECC-defining performance>.

NRF AI/Wiki-RAG version

Future materials development increasingly requires integration of literature knowledge, experimental data, and AI-assisted design. However, Physical AI for cementitious composites is premature without a source-grounded and continuously updated data infrastructure. The ECC/EGC literature contains rich evidence on binder chemistry, fiber bridging, flaw design, crack-width control, self-healing, and sustainability, but these data remain scattered across papers. A Wiki-RAG based knowledge system is therefore needed to organize verified claims, material variables, and performance metrics into a retrievable design foundation for AI-assisted mixture optimization.

6. Novelty paragraph template

Template

The novelty of this proposal lies in <specific mechanism/design/workflow>, rather than <generic or already established route>. Unlike prior studies that mainly focused on <prior route>, this project will <specific new action> by integrating <design variables / evidence layer / methods>. This will allow <expected scientific advance> and provide <practical or societal benefit>.

Example: low-fiber EGC

The novelty lies in reducing the strain-hardening threshold through intentional flaw and matrix design, rather than relying only on increased PE fiber volume. By treating EPS beads and lightweight inclusions as designed crack-activation features, the project will test whether low-fiber EGC can retain distributed cracking and tensile ductility at reduced fiber demand.

Example: Wiki-RAG proposal

The novelty lies in converting the ECC/EGC literature and Lee lab publication corpus into a source-grounded Wiki-RAG system for mixture reasoning. Unlike generic AI-based materials searches, the proposed system will preserve claim-evidence links, verification status, material variables, and performance metrics, allowing AI-generated mixture candidates to be constrained by experimentally grounded knowledge.

7. Methodology justification template

Template

The proposed methodology is justified by the Atlas evidence hierarchy. First, <global mechanism> is anchored in Victor Li's ECC framework and foundational PSH/fiber-bridging studies. Second, <extension evidence> shows that this mechanism can be extended to <target material system>. Third, Lee lab publications provide preliminary evidence that <lab-specific material route> is feasible. Therefore, the proposed experiments will focus on <method variables> and evaluate <performance metrics> using <test methods>.

Method selection guide

NRF AI/Wiki-RAG methodology version

The methodology will proceed in three layers. First, source notes, paper cards, and claim-evidence rows will be standardized as RAG-ready chunks with metadata for binder type, fiber type, flaw agent, test method, and verification status. Second, retrieval templates will route user questions to concept nodes, material-system nodes, lab positioning notes, and claim-evidence rows. Third, AI-generated mixture hypotheses will be treated as testable candidates, not conclusions, and will be validated through direct tensile testing, crack-width analysis, and durability or self-healing evaluation where relevant.

8. Risk and mitigation template

Table template

| Risk | Why it matters | Mitigation | Atlas evidence to cite |
|---|---|---|---|
| <risk> | <consequence> | <mitigation plan> | `<source path>` |

Common risks and mitigations

NRF AI/Wiki-RAG risk paragraph

A key risk is that generative AI may produce plausible but unsupported mixture recommendations. This will be mitigated by requiring every generated recommendation to cite retrieved Atlas evidence, identify its verification level, and specify experimental validation tests. Another risk is incomplete original PDF verification for some extension sources; therefore, source status will be preserved and claims will be downgraded when evidence is only source-note based.

9. Expected impact template

Scientific impact

Scientifically, the project will clarify how <mechanism> controls <performance> in <material system>. By linking micromechanical design variables to verified performance metrics, the work will extend the ECC/EGC knowledge base beyond isolated mixture studies.

Practical impact

Practically, the project will support the design of ductile, crack-width-controlled, and lower-carbon cementitious composites for resilient infrastructure. The proposed approach can reduce reliance on <high-carbon binder / virgin fiber / trial-and-error mixture design> while preserving performance.

AI/Wiki-RAG impact

The Wiki-RAG system will provide a reusable research infrastructure for source-grounded materials design. It will allow researchers and students to retrieve verified evidence, compare material systems, identify design gaps, and generate experimentally testable mixture hypotheses. This creates a realistic pathway toward future Physical AI by first establishing persistent, curated, and verifiable data foundations.

10. Citation grounding rule

Every proposal paragraph should be traceable to source evidence.

Required citation levels

Internal citation format

(Source: `relative/path/file.md`; evidence: `relative/path/matrix.csv`, paper_id=`...`)

No-evidence rule

If the Atlas has no evidence row for a claim, write:

The Atlas currently does not contain a verified evidence row for this specific claim.

or downgrade the claim:

This should be framed as a proposal hypothesis rather than an established finding.

Related workflow files

• 09_ai_workflows/wiki_rag_ingestion_plan.md
• 09_ai_workflows/atlas_query_templates.md
• 09_ai_workflows/reviewer_response_workflow.md
• 09_ai_workflows/original_pdf_matching_protocol.md
• 06_lab_position/materials_ai_wikirag_position.md
• 06_lab_position/our_lab_position_map.md
• 07_visualization/by_lee_lab_publications_claim_evidence_matrix.csv