Skip to content

Round 1 — Batch Summaries

45 agents read ~40 papers each, in parallel. Each agent produced a one-page summary: paper inventory, tags, cross-references, gaps. These batch summaries fed Round 2 (domain compression).

The files are the raw reading log — useful for auditing individual claims back to source papers.

How the corpus was partitioned

%%{init: {"theme":"base","themeVariables":{"fontSize":"15px","fontFamily":"Inter, system-ui, -apple-system, sans-serif","primaryTextColor":"#1a1a1a","lineColor":"#666"},"flowchart":{"nodeSpacing":25,"rankSpacing":30,"padding":14,"useMaxWidth":true}}}%%
flowchart TB
    Corpus[("<b>1,952 papers</b><br/>alphabetical sort<br/>literature_review/")]:::c

    subgraph Batches ["&nbsp;<b>9 batches · each ~200 papers</b>&nbsp;"]
        direction TB
        B1["<b>Batch 01</b> · files 1–200 · 1902–2018 mix"]:::b
        B2["<b>Batch 02</b> · files 201–400 · 2019–2020"]:::b
        B3["<b>Batch 03</b> · files 401–600 · 2020–2022"]:::b
        B4["<b>Batch 04</b> · files 601–800 · 2023–2024"]:::b
        B5["<b>Batch 05</b> · files 801–1000 · 2024 theses"]:::b
        B6["<b>Batch 06</b> · files 1001–1200 · 2024–2025 geotech"]:::b
        B7["<b>Batch 07</b> · files 1201–1400 · <b>DNV standards cluster</b>"]:::b
        B8["<b>Batch 08</b> · files 1401–1600 · 2025 SHM"]:::b
        B9["<b>Batch 09</b> · files 1601–1952 · 2025–2026 frontier"]:::b

        B1 --> B2 --> B3 --> B4 --> B5 --> B6 --> B7 --> B8 --> B9
    end

    Reading["<b>Each batch → 5 agents → 5 summaries</b><br/><span style='font-size:13px'>(45 reading tasks · ~40 papers each)</span>"]:::r

    Corpus ==> Batches ==> Reading

    classDef c fill:#f4f4f4,stroke:#444,stroke-width:2px,color:#222
    classDef b fill:#e3f2fd,stroke:#1565c0,stroke-width:1.5px,color:#0d47a1,text-align:left
    classDef r fill:#fff3e0,stroke:#e65100,stroke-width:2px,color:#e65100

    style Batches fill:#f0f7ff,stroke:#90caf9,stroke-dasharray:5 5

Papers per batch

%%{init: {"theme": "base"}}%%
pie showData
    title Paper count per batch (approximate)
    "Batch 01 · files 1–200" : 200
    "Batch 02 · files 201–400" : 200
    "Batch 03 · files 401–600" : 200
    "Batch 04 · files 601–800" : 200
    "Batch 05 · files 801–1000" : 200
    "Batch 06 · files 1001–1200" : 200
    "Batch 07 · files 1201–1400" : 200
    "Batch 08 · files 1401–1600" : 200
    "Batch 09 · files 1601–1952" : 352

Vintage distribution — years covered

%%{init: {"theme":"base","themeVariables":{"fontSize":"16px","fontFamily":"Inter, system-ui, -apple-system, sans-serif","primaryTextColor":"#1a1a1a","lineColor":"#666"},"flowchart":{"nodeSpacing":18,"rankSpacing":22,"padding":14,"useMaxWidth":false}}}%%
flowchart TB
    V1902["<b>1902 ·</b> Gibbs — statistical mechanics (stub)"]:::y
    V1940["<b>1940s–50s ·</b> von Karman · Feynman · Dirac — foundational physics"]:::y
    V1965["<b>1965 ·</b> Sokolovskii — granular media statics"]:::y
    V1972["<b>1972 ·</b> Vesic — cavity expansion theory"]:::y
    V1975["<b>1975–2001 ·</b> PhD thesis cluster — Morray · Randolph · Rowe · Houlsby · Byrne"]:::y
    V2010["<b>2010 ·</b> LeBlanc — cyclic stiffening"]:::y
    V2014["<b>2014 ·</b> DJ Kim tripod bucket centrifuge · Cox suction caisson cyclic"]:::y
    V2017["<b>2017 ·</b> PISA launch · Borkum Riffgrund 1 operational"]:::y
    V2020["<b>2020 ·</b> OWT-monitoring SHM papers proliferate"]:::y
    V2022["<b>2022 ·</b> Ciancimino scour rules · Stuyts monitored-f gap"]:::y
    V2024["<b>2024 ·</b> Large-diameter monopile extension · ML-heavy year"]:::y
    V2025["<b>2025 ·</b> Jin scour-capacity · Kim tripod centrifuge published"]:::y
    V2026["<b>2026 ·</b> Current frontier — Hong SDEE · Zhang sloping-bed · Bull ensemble DT"]:::y

    V1902 --> V1940 --> V1965 --> V1972 --> V1975 --> V2010
    V2010 --> V2014 --> V2017 --> V2020 --> V2022 --> V2024
    V2024 --> V2025 --> V2026

    classDef y fill:#fff3e0,stroke:#ef6c00,stroke-width:2px,color:#e65100,text-align:left

Batch × agent reading log

The table below is the reading log. Every row links to a standalone summary page.

# Batch Agent Scope File
1 01 1 Files 1–40 · 1902–2009 · foundational mechanics, offshore geotechnics, reliability, structural stability batch01_agent1.md
2 01 2 Files 41–80 batch01_agent2.md
3 01 3 Files 81–120 batch01_agent3.md
4 01 4 Files 121–160 · 2018 · offshore wind, geotech foundations, SHM, scour, fatigue batch01_agent4.md
5 01 5 Files 161–200 · 2018–2019 · offshore geotechnics, wind energy, structural engineering, data-driven methods batch01_agent5.md
6 02 1 Files 201–240 · 2019 batch02_agent1.md
7 02 2 Files 241–280 · late-2019 to early-2020 · OWT foundations, SSI, SHM, deep-learning PHM, centrifuge, geotech reliability batch02_agent2.md
8 02 3 Files 281–320 batch02_agent3.md
9 02 4 Files 321–360 · 2020 · offshore wind, geotech foundations, DL for geosciences batch02_agent4.md
10 02 5 Files 361–400 · 2020–2021 batch02_agent5.md
11 03 1 Files 401–440 batch03_agent1.md
12 03 2 Files 441–480 · 2021–2022 · OWT, geotech, SHM, ML, slope stability, foundations, sensing batch03_agent2.md
13 03 3 Files 481–520 batch03_agent3.md
14 03 4 Files 521–560 · 2022 batch03_agent4.md
15 03 5 Files 561–600 batch03_agent5.md
16 04 1 Files 601–640 batch04_agent1.md
17 04 2 Files 641–680 · 2023 batch04_agent2.md
18 04 3 Files 681–720 batch04_agent3.md
19 04 4 Files 721–760 · 2024 batch04_agent4.md
20 04 5 Files 761–800 · Simpson 2024 → Kang 2025 batch04_agent5.md
21 05 1 Files 801–840 batch05_agent1.md
22 05 2 Files 841–880 batch05_agent2.md
23 05 3 Files 881–920 batch05_agent3.md
24 05 4 Files 921–960 batch05_agent4.md
25 05 5 Files 961–1000 · Jawalageri 2022 → Malekjafarian 2020 batch05_agent5.md
26 06 1 Files 1001–1040 batch06_agent1.md
27 06 2 Files 1041–1080 batch06_agent2.md
28 06 3 Files 1081–1120 batch06_agent3.md
29 06 4 Files 1121–1160 batch06_agent4.md
30 06 5 Files 1161–1200 batch06_agent5.md
31 07 1 Files 1201–1240 batch07_agent1.md
32 07 2 Files 1241–1280 batch07_agent2.md
33 07 3 Files 1281–1320 batch07_agent3.md
34 07 4 Files 1321–1360 · DNV offshore standards and recommended practices (full set) batch07_agent4.md
35 07 5 Files 1361–1400 batch07_agent5.md
36 08 1 Files 1401–1440 batch08_agent1.md
37 08 2 Files 1441–1480 batch08_agent2.md
38 08 3 Files 1481–1520 batch08_agent3.md
39 08 4 Files 1521–1560 batch08_agent4.md
40 08 5 Files 1561–1600 batch08_agent5.md
41 09 1 Files 1601–1670 batch09_agent1.md
42 09 2 Files 1671–1740 batch09_agent2.md
43 09 3 Files 1741–1810 batch09_agent3.md
44 09 4 Files 1811–1880 batch09_agent4.md
45 09 5 Files 1881–1952 · final batch batch09_agent5.md

Notable batches to read first

If you want to sample the reading log rather than walk it sequentially, these five batches carry the highest density of dissertation-relevant content.

  • batch01_agent1 — Foundational mechanics

    1902–2009 · Gibbs statistical mechanics through 2001 Kuhn / Roth theses. The historical context for every modern claim in the portfolio.

    Read

  • batch01_agent4 — 2018 OWT + SHM boom

    The year vibration-based SHM on offshore wind matured. Prendergast, Weijtjens, Kim, and the early PISA papers all cluster here.

    Read

  • batch03_agent4 — 2022 state of the art

    Stuyts monitored-f gap, Ciancimino local-vs-general scour rules, Kariyawasam mode-shape work. The master-map debates crystallise here.

    Read

  • batch07_agent4 — DNV standards

    Every DNV OS and RP relevant to offshore wind geotechnics and SHM in a single agent's output. Indispensable reference for compliance claims.

    Read

  • batch09_agent5 — 2025–2026 frontier

    Hong 2026 SDEE tetrapod, Zhang sloping-bed scour, Bull ensemble DT, Jin scour-capacity. The reviewer-suggested and most recent papers.

    Read

How the numbering works

  • Papers were sorted alphabetically in literature_review/ and serialised by position (1–1,952).
  • 9 batches × 5 agents = 45 reading tasks, each responsible for ~40 papers.
  • One batch (batch 07 agent 4) landed entirely on DNV standards because they cluster alphabetically under "DNV-…".
  • Batch 09 agents received ~70 papers each because the final 352-file segment exceeded the 5 × 40 = 200 budget.

What to do with these files

  1. Trace a claim — the master map cites a paper; find which batch read it; read that agent's summary for the un-compressed take.
  2. Spot lost context — if a debate only shows up in one agent's batch but not in any domain map, that's a signal the Round 2 compression dropped something.
  3. Quarterly refresh — new papers go into the alphabetical position, a new agent slot reads them, and domain maps are updated quarterly.

Provenance — how a claim propagates back

%%{init: {"theme":"base","themeVariables":{"fontSize":"16px","fontFamily":"Inter, system-ui, -apple-system, sans-serif","primaryTextColor":"#1a1a1a","lineColor":"#666"},"flowchart":{"nodeSpacing":30,"rankSpacing":45,"padding":16,"useMaxWidth":true}}}%%
flowchart TB
    P["<b>Paper introduction</b><br/><span style='font-size:14px'>manuscript.qmd</span>"]:::p
    M["<b>Master map</b><br/><span style='font-size:14px'>MASTER_KNOWLEDGE_MAP.md</span>"]:::m
    D["<b>Domain map</b><br/><span style='font-size:14px'>DOMAIN_X.md</span>"]:::d
    B["<b>Batch summary</b><br/><span style='font-size:14px'>batchNN_agentM.md</span>"]:::b
    O["<b>Original paper</b><br/><span style='font-size:14px'>literature_review/*.md</span>"]:::o

    P == "cites" ==> M
    M == "synthesises" ==> D
    D == "compresses" ==> B
    B == "reads" ==> O

    classDef p fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px,color:#4a148c
    classDef m fill:#fff3e0,stroke:#e65100,stroke-width:2px,color:#e65100
    classDef d fill:#e8f5e9,stroke:#2e7d32,stroke-width:2px,color:#1b5e20
    classDef b fill:#e3f2fd,stroke:#1565c0,stroke-width:2px,color:#0d47a1
    classDef o fill:#fafafa,stroke:#444,stroke-width:2px,color:#333