feat(experiments): Phase C — experiment ledger, contact-zone templates, A/B language templates, cross-culture matrix

backend/glossa_lab/EXPERIMENT_LEDGER.md

@@ -0,0 +1,87 @@
+# Experiment Ledger
+
+Comprehensive catalog of all experiment graph phase files in the Glossa Lab research platform.
+
+## Summary
+
+| Phase File | Category | Status | Purpose | Recommendation |
+|---|---|---|---|---|
+| experiment_graph.py | structural_analysis | active | Core engine with ~50 built-in atomic nodes and graph execution runtime | keep |
+| experiment_graph_ctt.py | ctt | active | CTT nodes: sign role classifier, admissibility filter, holdout recall, compound constraint, anchored SA | keep |
+| experiment_graph_indus_evidence.py | misc | active | Evidence graph nodes for literature/claims management | keep |
+| experiment_graph_phase14.py | structural_analysis | active | Block entropy, conditional entropy, Zipf-Mandelbrot, MI decay, epistatic detection, language verdict | keep |
+| experiment_graph_phase15.py | structural_analysis | active | Long-tail validity, cipher self-test, multi-hypothesis ranker, deciphered mapping exporter | keep |
+| experiment_graph_phase_legacy.py | misc | legacy | Migration shims for Phase 16-19 standalone scripts | keep |
+| experiment_graph_phase20.py | structural_analysis | active | Length-stratified spectral, cluster archaeology, Ferrara OCR, Fuls classifier | keep |
+| experiment_graph_phase21.py | structural_analysis | active | Repetition collapser, site stratifier, numerical-weight regression | keep |
+| experiment_graph_phase22.py | contact_zone | active | CDLI Meluhha-mention corpus, Meluhhan persons, Indus seals at Mesopotamia | keep |
+| experiment_graph_phase23.py | contact_zone | active | Refined seal audit, strict PN extraction, bilingual readout test | keep |
+| experiment_graph_phase24.py | contact_zone | active | Laursen Table 1, seal sign upgrade, persons-v2, bipartite readout v2 | keep |
+| experiment_graph_phase25.py | sa_variant | active | Janabiyah readout, held-out test, period-stratified replication, Tamil-Brahmi | keep |
+| experiment_graph_phase26.py | sa_variant | active | Provenience-stratified SA, Bayesian decoder, expanded readout | keep |
+| experiment_graph_phase27.py | sa_variant | active | Reverse Janabiyah, Bayesian v2, iconographic anchors | keep |
+| experiment_graph_phase28.py | archaeological | active | CISI Vol 3 OCR, Mahadevan crosswalk, allograph-aware SA | keep |
+| experiment_graph_phase29.py | archaeological | active | Corpus 10x expansion: M77, ePSD2, Fuls, ICIT loaders | keep |
+| experiment_graph_phase30.py | sa_variant | active | Length-cohort reverse Janabiyah, permutation test, Dravidian syllable LM | keep |
+| experiment_graph_phase_misc_gaps.py | misc | legacy | Phases 44-47, 202, 209-215, 254-256 wrapped as subprocess runners | keep |
+| experiment_graph_phase48_55.py | sa_variant | active | Full Indus decipherment pipeline (GPU mandatory) | keep |
+| experiment_graph_phase56_61.py | sa_variant | active | Expanded Parpola SA, phonotactic falsification (GPU mandatory) | keep |
+| experiment_graph_phase62_66.py | sa_variant | active | Ensemble fix, phonotactic filter, morpheme boundary, Sanskrit SA | keep |
+| experiment_graph_phase67_73.py | sa_variant | active | Sanskrit norm, formula translation, M267 validation, parser | keep |
+| experiment_graph_phase74_80.py | lm_scoring | active | Grammar test, Levit, place formula, SA agreement, semantic cluster, DEDR | keep |
+| experiment_graph_phase81_87.py | sa_variant | active | M293 deep-dive, seal translation, formula lexicon, phonology | keep |
+| experiment_graph_phase88_90.py | misc | active | Literature mine, DEDR expansion to 120, scholarly translations | keep |
+| experiment_graph_phase91_100.py | sa_variant | active | Anchor-120 SA, M293 SA, trigram, grammar, full-corpus runs | keep |
+| experiment_graph_phase101_103.py | archaeological | active | M293 iconographic, PDF extraction, personal name lexicon | keep |
+| experiment_graph_phase104_109.py | sa_variant | active | OCR, name signs, name SA, Tamil-Brahmi check, phoneme exhaustion | keep |
+| experiment_graph_phase110_115.py | sa_variant | active | Targeted SA, allographs, grammar inference, seal translations | keep |
+| experiment_graph_phase116_121.py | sa_variant | active | SA recalibration, grammar LOW, site semantics, arXiv | keep |
+| experiment_graph_phase122_123.py | cross_language | active | Syllabic SA, Munda/BMAC substrate hypothesis | keep |
+| experiment_graph_phase124_125.py | structural_analysis | active | Fish-sign polysemy, Arthasastra mining | keep |
+| experiment_graph_phase126.py | archaeological | active | ICIT corpus integration and sign inventory alignment | keep |
+| experiment_graph_phase127.py | contact_zone | active | Gulf corpus analysis, Roif mining, fish-sign polysemy | keep |
+| experiment_graph_phase128_133.py | sa_variant | active | SA refinement and anchor injection cycles | keep |
+| experiment_graph_phase134_141.py | structural_analysis | active | Falsification suite, advancement testing, master scorecard | keep |
+| experiment_graph_phase142_145.py | sa_variant | active | SA anchor injection and refinement cycles | keep |
+| experiment_graph_phase146_155.py | sa_variant | active | SA parameter exploration and convergence testing | keep |
+| experiment_graph_phase156_165.py | sa_variant | active | Advanced SA with refined anchors | keep |
+| experiment_graph_phase166_168.py | lm_scoring | active | Sibilant validation, Meluhhan expansion, blocker SA | keep |
+| experiment_graph_phase169_170.py | structural_analysis | active | Master synthesis, grammar variance — computational frontier | keep |
+| experiment_graph_phase171_178.py | structural_analysis | active | Network centrality, betweenness stratification, network deep-dive | keep |
+| experiment_graph_phase179_180.py | misc | active | Literature mine, Mesopotamian contact mine | keep |
+| experiment_graph_phase181.py | archaeological | active | aDNA archaeogenetics mine | keep |
+| experiment_graph_phase182.py | misc | active | Deep evidence mine | keep |
+| experiment_graph_phase183.py | misc | active | Bulk mine 5000 (superseded by phase184) | keep |
+| experiment_graph_phase184.py | misc | active | Bulk mine 5000 second run, fresh clusters | keep |
+| experiment_graph_phase185_189.py | cross_language | active | Fish-sign battery, Elamo-Dravidian gap, commodity semantic, north Dravidian LM | keep |
+| experiment_graph_phase190_192.py | sa_variant | active | Elamite anchor injection, grammar validation | keep |
+| experiment_graph_phase193_195.py | sa_variant | active | SA rerun, SSRN fetch, grammar revalidation | keep |
+| experiment_graph_phase196_201.py | sa_variant | active | Mine3, top-8, DEDR lookup, triple-LM, inscription reading | keep |
+| experiment_graph_phase203_205.py | cross_language | active | E28 falsification, McAlpin cognates, Bayesian phylogenetics | keep |
+| experiment_graph_phase206_208.py | sa_variant | active | Anchor injection M692/M861, SA rerun 404, mine5 | keep |
+| experiment_graph_phase216_220.py | sa_variant | active | SA recalibration, site semantic, arXiv, Parpola/CISI | keep |
+| experiment_graph_phase221_225.py | sa_variant | active | P324/P122 investigation, CISI injection, slot mismatch | keep |
+| experiment_graph_phase226_228.py | sa_variant | active | P122 phonetic, P324 formula, CISI tripartite | keep |
+| experiment_graph_phase229.py | sa_variant | active | CISI anchor SA test, M122 upgrade | keep |
+| experiment_graph_phase230_234.py | contact_zone | active | Cross-ref matrix, indirect bilingual, cultural/demographic | keep |
+| experiment_graph_phase235_236.py | cross_language | active | Elamite–PDr bridge, Sanskrit loanword mapping | keep |
+| experiment_graph_phase237_246.py | sa_variant | active | Blocker mine, batch upgrades, synthesis, SA crossing | keep |
+| experiment_graph_phase248_253.py | sa_variant | active | Ceiling-breaker mine, allograph, semantic constraint | keep |
+| experiment_graph_phase257_294.py | sa_variant | active | SA reruns, Yajnadevam, DEDR resolution, 605/605 decipherment | keep |
+| experiment_graph_phase295_297.py | misc | active | Bulk mine May 2026, cross-reference, gap analysis | keep |
+| experiment_graph_phase298_308.py | cross_language | active | Munda mine/SA, substrate, archaeology, DEDR, Elamite baseline | keep |
+| experiment_graph_phase322_362.py | sa_variant | active | May 2026 decipherment advancement session | keep |
+| experiment_graph_contact_zone.py | contact_zone | active | KL divergence, synthesis, A/B comparison for contact signals | keep |
+| experiment_graph_ab_language.py | cross_language | active | A/B language SA: Dravidian vs Sanskrit/Munda/Hebrew, LM consistency | keep |
+| experiment_graph_cross_culture.py | cross_language | active | Cultural contact matrix, script family classifier | keep |
+
+## Category Legend
+
+- **structural_analysis** — Entropy, positional, spectral, network analysis of sign/symbol systems
+- **lm_scoring** — Language model scoring, grammar tests, validation
+- **sa_variant** — Simulated annealing decipherment runs and variants
+- **ctt** — Constraint Topology Theory framework
+- **contact_zone** — Mesopotamian contact corpus, bilingual analysis, seals
+- **cross_language** — Cross-language comparison, substrate hypothesis, phylogenetics
+- **archaeological** — OCR, corpus expansion, archaeogenetics, iconography
+- **misc** — Literature mining, evidence management, infrastructure, legacy shims

backend/glossa_lab/api/experiments.py

@@ -38,6 +38,17 @@
 router = APIRouter()
 
 
+@router.get("/experiments/metadata")
+async def get_experiments_metadata() -> list[dict[str, Any]]:
+    """Return experiment ledger metadata for all registered nodes.
+
+    Merges the static experiment_ledger.json with live ATOMIC_NODES
+    registration data. Used by the frontend ExperimentRegistry component.
+    """
+    from glossa_lab.experiment_graph import get_experiment_metadata  # noqa: PLC0415
+    return get_experiment_metadata()
+
+
 @router.get("/experiments")
 async def list_experiments() -> list[dict[str, Any]]:
     """Return graph experiments only (H16 compliance).

backend/glossa_lab/experiment_graph.py

@@ -2834,6 +2834,39 @@ def _experiment_wrapper(inputs: dict, params: dict) -> dict:
 except Exception as _p322390_exc:  # noqa: BLE001
     logger.warning("Phase-322-390 nodes not registered: %s", _p322390_exc)
 
+# ── Contact-zone analysis templates (KL divergence, synthesis, A/B) ─────────
+try:
+    from glossa_lab.experiment_graph_contact_zone import (
+        _contact_zone_node_defs as _cz_defs,
+    )
+    for _d in _cz_defs():
+        ATOMIC_NODES[_d.id] = _d
+    logger.info("Registered %d contact-zone template nodes", len(_cz_defs()))
+except Exception as _cz_exc:  # noqa: BLE001
+    logger.warning("Contact-zone template nodes not registered: %s", _cz_exc)
+
+# ── A/B language comparison templates (anchored SA, consistency matrix) ──────
+try:
+    from glossa_lab.experiment_graph_ab_language import (
+        _ab_language_node_defs as _ab_defs,
+    )
+    for _d in _ab_defs():
+        ATOMIC_NODES[_d.id] = _d
+    logger.info("Registered %d A/B language template nodes", len(_ab_defs()))
+except Exception as _ab_exc:  # noqa: BLE001
+    logger.warning("A/B language template nodes not registered: %s", _ab_exc)
+
+# ── Cross-culture contact matrix + script family classifier ─────────────────
+try:
+    from glossa_lab.experiment_graph_cross_culture import (
+        _cross_culture_node_defs as _cc_defs,
+    )
+    for _d in _cc_defs():
+        ATOMIC_NODES[_d.id] = _d
+    logger.info("Registered %d cross-culture template nodes", len(_cc_defs()))
+except Exception as _cc_exc:  # noqa: BLE001
+    logger.warning("Cross-culture template nodes not registered: %s", _cc_exc)
+
 # ── Research Loop Runner (meta-node for Experiment Builder)
 try:
     from glossa_lab.pipelines.research_loop import ResearchLoop as _RL
@@ -2869,6 +2902,61 @@ def _research_loop_runner(inputs: dict, params: dict) -> dict:
     logger.warning("ResearchLoopRunner not registered: %s", _rl_exc)
 
 
+# ── Experiment metadata (ledger-backed) ──────────────────────────────────────
+
+def get_experiment_metadata() -> list[dict[str, Any]]:
+    """Return ledger metadata for all registered experiment nodes.
+
+    Merges the static experiment_ledger.json with live registration data
+    from ATOMIC_NODES to provide a unified metadata view.
+    """
+    ledger_path = Path(__file__).parent / "experiment_ledger.json"
+    ledger: list[dict] = []
+    if ledger_path.exists():
+        try:
+            ledger = json.loads(ledger_path.read_text("utf-8"))
+        except Exception:  # noqa: BLE001
+            pass
+
+    # Build a lookup from node ID to ledger entry
+    node_to_ledger: dict[str, dict] = {}
+    for entry in ledger:
+        for node_name in entry.get("key_nodes", []):
+            node_to_ledger[node_name] = entry
+
+    # Merge live registration with ledger
+    result: list[dict[str, Any]] = []
+    for node_id, node_def in ATOMIC_NODES.items():
+        ledger_entry = node_to_ledger.get(node_id, {})
+        result.append({
+            "id": node_id,
+            "display_name": node_def.name,
+            "category": ledger_entry.get("category", node_def.category),
+            "phase": ledger_entry.get("phases", ""),
+            "description": node_def.description,
+            "status": ledger_entry.get("status", "active"),
+            "superseded_by": ledger_entry.get("superseded_by"),
+            "source_file": ledger_entry.get("file", ""),
+        })
+
+    # Also include ledger-only entries (files without individually named nodes)
+    seen_files = {e.get("source_file") for e in result if e.get("source_file")}
+    for entry in ledger:
+        if entry.get("file") not in seen_files:
+            result.append({
+                "id": entry.get("file", "").replace(".py", ""),
+                "display_name": entry.get("file", "").replace("experiment_graph_", "").replace(".py", ""),
+                "category": entry.get("category", "misc"),
+                "phase": entry.get("phases", ""),
+                "description": entry.get("purpose", ""),
+                "status": entry.get("status", "active"),
+                "superseded_by": entry.get("superseded_by"),
+                "source_file": entry.get("file", ""),
+            })
+
+    return result
+
+
 # ── Graph execution
 
 def _topo_sort(nodes: list[dict], edges: list[dict]) -> list[dict]: