Add negative act form experiment

docs/experiments.md

@@ -1890,6 +1890,78 @@ cross-pattern reconciliation is justified.
 Artifacts are preserved under
 `artifacts/experiments/explicit_rejection_gold_v0/20260820_qwen35_9b_single_run/`.
 
+## EXP-0035 — Negative Act Form V0
+
+Status: Experimental; successful for form classification with normalization
+limitations
+
+Date: 2026-08-20
+
+EXP-0034 failed because the binary `explicit_action_rejection | none` question
+collapsed materially different negative acts. It missed self-contained
+non-pursuit and promoted personal preference, recommendation and temporary
+non-action to rejection. This isolated follow-up tested only whether those
+evidence-near forms can be distinguished before any normative derivation. It
+does not derive rejection, decision, outcome, topic closure, responsibility or
+protocol status, and EXP-0034 remained unchanged.
+
+The strict output schema contains exactly `observation_id`,
+`negative_act_form` and `normalized_action_text`. The closed form vocabulary is
+`explicit_non_pursuit`, `personal_preference`, `recommendation`,
+`temporary_non_action` and `none`. Non-`none` forms require non-empty normalized
+action text; `none` requires null. Rejection, status, decision, outcome,
+responsibility and other normative fields are forbidden recursively. Local
+context may resolve a candidate observation's pronoun, but the schema contains
+no target relation and the experiment exposes no derivation function.
+
+Gold results:
+
+- NA-01 explicit non-pursuit: PARTIAL. The form was correct; `working with Dr.
+  Schlummer` omitted the continuation aspect from normalization.
+- NA-02 paraphrased explicit non-pursuit: PASS.
+- NA-03 personal preference: PARTIAL. The form was correct, but normalization
+  repeated `Ich würde das nicht machen` instead of resolving the real-plant
+  trial target.
+- NA-04 negative recommendation: PARTIAL. The form was correct; the normalized
+  English action used the loose rendering `real asset` for `reale Anlage`.
+- NA-05 temporary non-action: PASS.
+- NA-06 concern only: PASS with `none` and null action text.
+- NA-07 uncertainty: PASS with `none` and null action text.
+- NA-08 factual negation: PASS with `none` and null action text.
+
+Expected-versus-actual form confusion was entirely diagonal:
+
+| Expected form | Actual form | Count |
+| --- | --- | ---: |
+| `explicit_non_pursuit` | `explicit_non_pursuit` | 2 |
+| `personal_preference` | `personal_preference` | 1 |
+| `recommendation` | `recommendation` | 1 |
+| `temporary_non_action` | `temporary_non_action` | 1 |
+| `none` | `none` | 3 |
+
+Configuration: exactly eight successful sequential `qwen3.5:9B` calls, one
+per case, temperature 0, `think=false`, `num_ctx=16384`,
+`num_predict=1024`, no retries, no voting and no prompt changes. There were zero
+technical failures. Aggregate runner time was 8.688 seconds; summed per-call
+time was 8.686 seconds, with 4,183 prompt-evaluation tokens and 310 evaluation
+tokens.
+
+The result was five PASS, three PARTIAL and zero FAIL. All eight
+`negative_act_form` classifications matched Gold. There was no unsupported
+semantic strengthening and no rejection, status, decision, outcome,
+responsibility or topic-closure leakage. Normalized action meaning was fully
+acceptable in five cases and imperfect in three.
+
+Conclusion: the finer evidence-near form vocabulary successfully distinguished
+the four semantic boundaries that defeated the binary rejection experiment in
+this small Gold set. The result supports separating negative-act-form
+recognition from later normative derivation, but local target normalization is
+not yet uniformly reliable. It does not justify modifying EXP-0034, deriving
+rejection, production integration or beginning cross-pattern reconciliation.
+
+Artifacts are preserved under
+`artifacts/experiments/negative_act_form_v0/20260820_qwen35_9b_single_run/`.
+
 ## EXP-0026 — Topic-oriented Discussion Subject reconstruction V2 prototype
 
 Date: 2026-08-11

scripts/run_negative_act_form_experiment.py

@@ -0,0 +1,16 @@
+#!/usr/bin/env python3
+"""Repository entry point for the Negative Act Form experiment."""
+
+import sys
+from pathlib import Path
+
+
+REPO_ROOT = Path(__file__).resolve().parents[1]
+if str(REPO_ROOT) not in sys.path:
+    sys.path.insert(0, str(REPO_ROOT))
+
+from src.meeting_lab.controlled_semantic_derivation.experiment_negative_act import main  # noqa: E402
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

src/meeting_lab/controlled_semantic_derivation/experiment_negative_act.py

@@ -0,0 +1,277 @@
+#!/usr/bin/env python3
+"""Isolated evidence-near Negative Act Form classification experiment."""
+
+from __future__ import annotations
+
+import argparse
+import json
+import time
+from pathlib import Path
+from typing import Any
+
+from .experiment_h import (
+    DEFAULT_ENDPOINT,
+    DEFAULT_MODEL,
+    DerivationValidationError,
+    OBSERVATION_KEYS,
+    build_ollama_payload,
+    call_ollama,
+)
+
+
+GOLD_SCHEMA_VERSION = "experimental-negative-act-form-gold-v0"
+RECOGNITION_KEYS = {"observation_id", "negative_act_form", "normalized_action_text"}
+NEGATIVE_ACT_FORMS = {
+    "explicit_non_pursuit", "personal_preference", "recommendation",
+    "temporary_non_action", "none",
+}
+FORBIDDEN_LLM_KEYS = {
+    "rejection_form", "explicitly_rejected", "status", "decision", "outcome",
+    "topic_status", "responsible_person", "responsibility", "owner",
+    "requested_actor", "action_item", "protocol", "protocol_category",
+    "confidence", "relation", "relations", "graph", "unresolved_issue",
+}
+
+PROMPT_TEMPLATE = """Classify only the negative semantic form expressed by the candidate observation, using earlier supplied V3-style observations only as local context for pronouns or shortened references.
+
+The candidate observation is {candidate_observation_id}.
+
+Choose exactly one negative_act_form:
+- explicit_non_pursuit: explicitly states that an action, option, collaboration, or course will not be continued or pursued. This is stronger than preference, advice, or temporary delay.
+- personal_preference: the speaker states what they personally would or would not do, without establishing collective non-pursuit.
+- recommendation: the speaker advises for or against an action without establishing abandonment.
+- temporary_non_action: the action is postponed, deferred, or explicitly not done for now without abandonment.
+- none: none of those four forms is present, including mere concern, uncertainty, negative sentiment, or factual negation.
+
+Do not collapse non-pursuit into temporary non-action. Do not convert a personal conditional preference into collective non-pursuit. Do not convert advice into non-pursuit. Speaker identity does not change personal preference into collective non-pursuit.
+
+When the form is not none, return concise normalized action meaning. Resolve a pronoun only from the supplied local context. If its target is genuinely ambiguous, return none rather than guessing. When the form is none, normalized_action_text must be null. Keep normalized action text in the observation language.
+
+Do not derive or output rejection, status, decision, outcome, topic closure, responsibility, ownership, Action Item, protocol category, confidence, relations, graphs, or unresolved issues.
+
+Return exactly this JSON shape and no additional fields:
+{{
+  "observation_id": "{candidate_observation_id}",
+  "negative_act_form": "explicit_non_pursuit | personal_preference | recommendation | temporary_non_action | none",
+  "normalized_action_text": "concise action meaning" | null
+}}
+
+V3-style observations:
+{observations_json}
+"""
+
+
+def _exact_keys(value: dict[str, Any], required: set[str], location: str) -> None:
+    missing = required - value.keys()
+    unknown = value.keys() - required
+    if missing:
+        raise DerivationValidationError(f"{location} missing required keys: {sorted(missing)}")
+    if unknown:
+        raise DerivationValidationError(f"{location} has unknown keys: {sorted(unknown)}")
+
+
+def _nonempty_text(value: Any, location: str) -> str:
+    if not isinstance(value, str) or not value.strip():
+        raise DerivationValidationError(f"{location} must be a non-empty string")
+    return value.strip()
+
+
+def _validate_observations(observations: Any) -> None:
+    if not isinstance(observations, list) or not observations:
+        raise DerivationValidationError("observations must be a non-empty list")
+    seen_observations: set[str] = set()
+    seen_evidence: set[str] = set()
+    for index, observation in enumerate(observations):
+        location = f"observations[{index}]"
+        if not isinstance(observation, dict):
+            raise DerivationValidationError(f"{location} must be an object")
+        _exact_keys(observation, OBSERVATION_KEYS, location)
+        observation_id = _nonempty_text(observation["observation_id"], f"{location}.observation_id")
+        evidence_id = _nonempty_text(observation["evidence_id"], f"{location}.evidence_id")
+        if observation_id in seen_observations or evidence_id in seen_evidence:
+            raise DerivationValidationError("observation and evidence provenance must be unique")
+        seen_observations.add(observation_id)
+        seen_evidence.add(evidence_id)
+        _nonempty_text(observation["content"], f"{location}.content")
+        _nonempty_text(observation["speaker"], f"{location}.speaker")
+        for field in ("named_person", "addressee"):
+            if observation[field] is not None:
+                _nonempty_text(observation[field], f"{location}.{field}")
+
+
+def load_gold_cases(path: Path) -> list[dict[str, Any]]:
+    data = json.loads(path.read_text(encoding="utf-8-sig"))
+    if not isinstance(data, dict):
+        raise DerivationValidationError("Gold fixture must be an object")
+    _exact_keys(data, {"schema_version", "cases"}, "Gold fixture")
+    if data["schema_version"] != GOLD_SCHEMA_VERSION:
+        raise DerivationValidationError("unexpected Gold fixture schema_version")
+    cases = data["cases"]
+    if not isinstance(cases, list) or not cases:
+        raise DerivationValidationError("Gold fixture cases must be a non-empty list")
+    seen: set[str] = set()
+    for case in cases:
+        _exact_keys(case, {"case_id", "description", "observations", "expected"}, "Gold case")
+        case_id = _nonempty_text(case["case_id"], "Gold case.case_id")
+        if case_id in seen:
+            raise DerivationValidationError(f"duplicate case ID: {case_id}")
+        seen.add(case_id)
+        _validate_observations(case["observations"])
+        if len(case["observations"]) not in (1, 2):
+            raise DerivationValidationError("Negative Act cases require one or two observations")
+    return cases
+
+
+def build_prompt(case: dict[str, Any]) -> str:
+    observations = case["observations"]
+    _validate_observations(observations)
+    candidate_id = observations[-1]["observation_id"]
+    return PROMPT_TEMPLATE.format(
+        candidate_observation_id=candidate_id,
+        observations_json=json.dumps(observations, ensure_ascii=False, indent=2),
+    )
+
+
+def parse_model_json(raw_text: str) -> dict[str, Any]:
+    data = json.loads(raw_text)
+    if not isinstance(data, dict):
+        raise DerivationValidationError("semantic classification must be an object")
+    return data
+
+
+def _reject_forbidden_keys(value: Any, location: str = "output") -> None:
+    if isinstance(value, dict):
+        forbidden = FORBIDDEN_LLM_KEYS.intersection(value)
+        if forbidden:
+            raise DerivationValidationError(f"{location} contains forbidden semantic keys: {sorted(forbidden)}")
+        for key, item in value.items():
+            _reject_forbidden_keys(item, f"{location}.{key}")
+    elif isinstance(value, list):
+        for index, item in enumerate(value):
+            _reject_forbidden_keys(item, f"{location}[{index}]")
+
+
+def validate_classification(data: Any, observations: list[dict[str, Any]]) -> dict[str, Any]:
+    _validate_observations(observations)
+    if not isinstance(data, dict):
+        raise DerivationValidationError("semantic classification must be an object")
+    _reject_forbidden_keys(data)
+    _exact_keys(data, RECOGNITION_KEYS, "output")
+    observation_id = _nonempty_text(data["observation_id"], "output.observation_id")
+    if observation_id not in {item["observation_id"] for item in observations}:
+        raise DerivationValidationError("classification references unknown observation")
+    form = data["negative_act_form"]
+    if form not in NEGATIVE_ACT_FORMS:
+        raise DerivationValidationError("negative_act_form has an unsupported value")
+    action_text = data["normalized_action_text"]
+    if form == "none":
+        if action_text is not None:
+            raise DerivationValidationError("none form requires null normalized_action_text")
+    else:
+        _nonempty_text(action_text, "output.normalized_action_text")
+    return data
+
+
+def _concepts_present(text: str | None, concepts: list[list[str]]) -> bool:
+    if not concepts:
+        return text is None
+    if not isinstance(text, str):
+        return False
+    folded = text.casefold()
+    return all(any(alias.casefold() in folded for alias in alternatives) for alternatives in concepts)
+
+
+def evaluate_case(case: dict[str, Any], classification: dict[str, Any]) -> dict[str, Any]:
+    validate_classification(classification, case["observations"])
+    expected = case["expected"]
+    observation_correct = classification["observation_id"] == expected["observation_id"]
+    form_correct = classification["negative_act_form"] == expected["negative_act_form"]
+    action_correct = _concepts_present(classification["normalized_action_text"], expected["action_concepts"])
+    unsupported_strengthening = expected["negative_act_form"] == "none" and classification["negative_act_form"] != "none"
+    classification_label = "PASS" if observation_correct and form_correct and action_correct else ("PARTIAL" if observation_correct and form_correct else "FAIL")
+    return {
+        "case_id": case["case_id"], "classification": classification_label,
+        "expected_negative_act_form": expected["negative_act_form"],
+        "actual_negative_act_form": classification["negative_act_form"],
+        "observation_id_correct": observation_correct,
+        "normalized_action_meaning_correct": action_correct,
+        "unsupported_semantic_strengthening": unsupported_strengthening,
+        "normative_leakage": False,
+    }
+
+
+def _write_json(path: Path, value: Any) -> None:
+    path.write_text(json.dumps(value, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
+
+
+def run_experiment(args: argparse.Namespace) -> dict[str, Any]:
+    cases = load_gold_cases(args.cases)
+    args.output.mkdir(parents=True, exist_ok=False)
+    _write_json(args.output / "gold_cases.json", {"schema_version": GOLD_SCHEMA_VERSION, "cases": cases})
+    evaluations: list[dict[str, Any]] = []
+    successful_calls = 0
+    technical_failures = 0
+    started = time.perf_counter()
+    for case in cases:
+        case_dir = args.output / case["case_id"].lower()
+        case_dir.mkdir()
+        observations = case["observations"]
+        _write_json(case_dir / "v3_style_input_observations.json", observations)
+        prompt = build_prompt(case)
+        (case_dir / "prompt.txt").write_text(prompt, encoding="utf-8")
+        try:
+            raw, metadata = call_ollama(args.endpoint, args.model, prompt, args.timeout, args.num_ctx, args.num_predict)
+            successful_calls += 1
+        except Exception as exc:  # one recorded attempt; never retry
+            technical_failures += 1
+            failure = {"case_id": case["case_id"], "classification": "FAIL", "technical_failure": True, "error_type": type(exc).__name__, "error": str(exc)}
+            _write_json(case_dir / "ollama_metadata.json", {"model": args.model, "configuration": {"temperature": 0, "think": False, "num_ctx": args.num_ctx, "num_predict": args.num_predict, "retries": 0}, "technical_failure": failure})
+            _write_json(case_dir / "structural_validation.json", {"valid": False, "error": str(exc)})
+            _write_json(case_dir / "evaluation.json", failure)
+            evaluations.append(failure)
+            continue
+        (case_dir / "raw_model_response.txt").write_text(raw + "\n", encoding="utf-8")
+        _write_json(case_dir / "ollama_metadata.json", metadata)
+        try:
+            parsed = parse_model_json(raw)
+            _write_json(case_dir / "parsed_semantic_classification.json", parsed)
+            evaluation = evaluate_case(case, parsed)
+            validation = {"valid": True, "error": None}
+        except (DerivationValidationError, json.JSONDecodeError) as exc:
+            validation = {"valid": False, "error_type": type(exc).__name__, "error": str(exc)}
+            evaluation = {"case_id": case["case_id"], "classification": "FAIL", "error": str(exc), "normative_leakage": "forbidden" in str(exc)}
+        _write_json(case_dir / "structural_validation.json", validation)
+        _write_json(case_dir / "evaluation.json", evaluation)
+        evaluations.append(evaluation)
+    summary = {
+        "experiment": "negative_act_form_v0", "model": args.model,
+        "successful_llm_call_count": successful_calls,
+        "technical_failed_call_count": technical_failures,
+        "runtime_seconds": round(time.perf_counter() - started, 3),
+        "counts": {label: sum(item["classification"] == label for item in evaluations) for label in ("PASS", "PARTIAL", "FAIL")},
+        "evaluations": evaluations,
+    }
+    _write_json(args.output / "summary.json", summary)
+    return summary
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(description="Run isolated Negative Act Form experiment")
+    parser.add_argument("cases", type=Path)
+    parser.add_argument("-o", "--output", type=Path, required=True)
+    parser.add_argument("--model", default=DEFAULT_MODEL)
+    parser.add_argument("--endpoint", default=DEFAULT_ENDPOINT)
+    parser.add_argument("--timeout", type=int, default=300)
+    parser.add_argument("--num-ctx", type=int, default=16384)
+    parser.add_argument("--num-predict", type=int, default=1024)
+    return parser.parse_args()
+
+
+def main() -> int:
+    summary = run_experiment(parse_args())
+    print(json.dumps(summary, ensure_ascii=False, indent=2))
+    return 0 if summary["counts"]["FAIL"] == 0 and summary["technical_failed_call_count"] == 0 else 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

tests/gold/negative_act_form_v0/cases.json

@@ -0,0 +1,66 @@
+{
+  "schema_version": "experimental-negative-act-form-gold-v0",
+  "cases": [
+    {
+      "case_id": "NA-01", "description": "Explicit non-pursuit",
+      "observations": [
+        {"observation_id": "obs_1", "evidence_id": "e1", "content": "Martin: Mit Dr. Schlummer arbeiten wir nicht weiter.", "speaker": "Martin", "named_person": "Dr. Schlummer", "addressee": null}
+      ],
+      "expected": {"observation_id": "obs_1", "negative_act_form": "explicit_non_pursuit", "action_concepts": [["schlummer"], ["arbeit", "collabor"], ["weiter", "fortsetz", "continu"]]}
+    },
+    {
+      "case_id": "NA-02", "description": "Explicit non-pursuit paraphrase",
+      "observations": [
+        {"observation_id": "obs_1", "evidence_id": "e1", "content": "Martin: Die externe Lösung verfolgen wir nicht weiter.", "speaker": "Martin", "named_person": null, "addressee": null}
+      ],
+      "expected": {"observation_id": "obs_1", "negative_act_form": "explicit_non_pursuit", "action_concepts": [["extern"], ["lösung", "solution"], ["weiter", "pursu", "continu"]]}
+    },
+    {
+      "case_id": "NA-03", "description": "Personal preference with local context",
+      "observations": [
+        {"observation_id": "obs_1", "evidence_id": "e1", "content": "Martin: Wir könnten die reale Anlage für den Versuch nutzen.", "speaker": "Martin", "named_person": null, "addressee": null},
+        {"observation_id": "obs_2", "evidence_id": "e2", "content": "Martin: Ich würde das nicht machen.", "speaker": "Martin", "named_person": null, "addressee": null}
+      ],
+      "expected": {"observation_id": "obs_2", "negative_act_form": "personal_preference", "action_concepts": [["real"], ["anlage", "plant"], ["versuch", "trial", "test"]]}
+    },
+    {
+      "case_id": "NA-04", "description": "Negative recommendation",
+      "observations": [
+        {"observation_id": "obs_1", "evidence_id": "e1", "content": "Martin: Wir könnten die reale Anlage verwenden.", "speaker": "Martin", "named_person": null, "addressee": null},
+        {"observation_id": "obs_2", "evidence_id": "e2", "content": "Martin: Ich würde eher davon abraten.", "speaker": "Martin", "named_person": null, "addressee": null}
+      ],
+      "expected": {"observation_id": "obs_2", "negative_act_form": "recommendation", "action_concepts": [["real"], ["anlage", "plant"], ["verwend", "use"]]}
+    },
+    {
+      "case_id": "NA-05", "description": "Temporary non-action",
+      "observations": [
+        {"observation_id": "obs_1", "evidence_id": "e1", "content": "Martin: Wir könnten die Waschstufe einbauen.", "speaker": "Martin", "named_person": null, "addressee": null},
+        {"observation_id": "obs_2", "evidence_id": "e2", "content": "Martin: Das machen wir erstmal noch nicht.", "speaker": "Martin", "named_person": null, "addressee": null}
+      ],
+      "expected": {"observation_id": "obs_2", "negative_act_form": "temporary_non_action", "action_concepts": [["waschstufe", "washing stage"], ["einbau", "install"]]}
+    },
+    {
+      "case_id": "NA-06", "description": "Concern only",
+      "observations": [
+        {"observation_id": "obs_1", "evidence_id": "e1", "content": "Martin: Wir könnten das neue Material einsetzen.", "speaker": "Martin", "named_person": null, "addressee": null},
+        {"observation_id": "obs_2", "evidence_id": "e2", "content": "Martin: Das wäre kritisch.", "speaker": "Martin", "named_person": null, "addressee": null}
+      ],
+      "expected": {"observation_id": "obs_2", "negative_act_form": "none", "action_concepts": []}
+    },
+    {
+      "case_id": "NA-07", "description": "Uncertainty",
+      "observations": [
+        {"observation_id": "obs_1", "evidence_id": "e1", "content": "Martin: Eine Möglichkeit wäre, die Waschstufe einzubauen.", "speaker": "Martin", "named_person": null, "addressee": null},
+        {"observation_id": "obs_2", "evidence_id": "e2", "content": "Martin: Ich weiß nicht, ob das sinnvoll ist.", "speaker": "Martin", "named_person": null, "addressee": null}
+      ],
+      "expected": {"observation_id": "obs_2", "negative_act_form": "none", "action_concepts": []}
+    },
+    {
+      "case_id": "NA-08", "description": "Factual negation",
+      "observations": [
+        {"observation_id": "obs_1", "evidence_id": "e1", "content": "Martin: Das Material ist nicht verfügbar.", "speaker": "Martin", "named_person": null, "addressee": null}
+      ],
+      "expected": {"observation_id": "obs_1", "negative_act_form": "none", "action_concepts": []}
+    }
+  ]
+}

tests/test_negative_act_form_experiment.py

@@ -0,0 +1,162 @@
+import argparse
+import json
+import tempfile
+import unittest
+from copy import deepcopy
+from pathlib import Path
+from unittest.mock import patch
+
+import src.meeting_lab.controlled_semantic_derivation.experiment_negative_act as module
+from src.meeting_lab.controlled_semantic_derivation.experiment_negative_act import (
+    DerivationValidationError,
+    build_ollama_payload,
+    build_prompt,
+    evaluate_case,
+    load_gold_cases,
+    parse_model_json,
+    run_experiment,
+    validate_classification,
+)
+
+
+GOLD_PATH = Path("tests/gold/negative_act_form_v0/cases.json")
+
+
+FORM_TEXT = {
+    "NA-01": "Zusammenarbeit mit Dr. Schlummer fortsetzen",
+    "NA-02": "externe Lösung weiterverfolgen",
+    "NA-03": "reale Anlage für den Versuch nutzen",
+    "NA-04": "reale Anlage verwenden",
+    "NA-05": "Waschstufe einbauen",
+}
+
+
+def classification_for(case):
+    expected = case["expected"]
+    return {
+        "observation_id": expected["observation_id"],
+        "negative_act_form": expected["negative_act_form"],
+        "normalized_action_text": FORM_TEXT.get(case["case_id"]),
+    }
+
+
+class NegativeActFormExperimentTests(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls):
+        cls.cases = load_gold_cases(GOLD_PATH)
+        cls.by_id = {case["case_id"]: case for case in cls.cases}
+
+    def test_fixture_contains_exactly_na_01_through_na_08(self):
+        self.assertEqual(list(self.by_id), [f"NA-{number:02d}" for number in range(1, 9)])
+
+    def test_exact_schema_is_accepted(self):
+        case = self.by_id["NA-01"]
+        self.assertEqual(validate_classification(classification_for(case), case["observations"]), classification_for(case))
+
+    def test_unknown_field_is_rejected(self):
+        case = self.by_id["NA-01"]
+        classification = classification_for(case)
+        classification["explanation"] = "extra"
+        with self.assertRaisesRegex(DerivationValidationError, "unknown keys"):
+            validate_classification(classification, case["observations"])
+
+    def test_invalid_enum_is_rejected(self):
+        case = self.by_id["NA-01"]
+        classification = classification_for(case)
+        classification["negative_act_form"] = "rejection"
+        with self.assertRaisesRegex(DerivationValidationError, "unsupported value"):
+            validate_classification(classification, case["observations"])
+
+    def test_non_none_requires_normalized_action_text(self):
+        case = self.by_id["NA-01"]
+        for value in (None, ""):
+            classification = classification_for(case)
+            classification["normalized_action_text"] = value
+            with self.subTest(value=value), self.assertRaises(DerivationValidationError):
+                validate_classification(classification, case["observations"])
+
+    def test_none_requires_null_normalized_action_text(self):
+        case = self.by_id["NA-06"]
+        classification = classification_for(case)
+        self.assertIsNone(classification["normalized_action_text"])
+        classification["normalized_action_text"] = "Material einsetzen"
+        with self.assertRaisesRegex(DerivationValidationError, "requires null"):
+            validate_classification(classification, case["observations"])
+
+    def test_forbidden_normative_fields_are_rejected_recursively(self):
+        case = self.by_id["NA-01"]
+        fields = (
+            "rejection_form", "explicitly_rejected", "status", "decision", "outcome",
+            "topic_status", "responsible_person", "responsibility", "owner",
+            "requested_actor", "action_item", "protocol_category", "confidence",
+            "relation", "relations", "graph", "unresolved_issue",
+        )
+        for field in fields:
+            classification = classification_for(case)
+            classification["wrapper"] = {field: "forbidden"}
+            with self.subTest(field=field), self.assertRaisesRegex(DerivationValidationError, "forbidden semantic keys"):
+                validate_classification(classification, case["observations"])
+
+    def test_unknown_observation_id_is_rejected(self):
+        case = self.by_id["NA-01"]
+        classification = classification_for(case)
+        classification["observation_id"] = "obs_99"
+        with self.assertRaisesRegex(DerivationValidationError, "unknown observation"):
+            validate_classification(classification, case["observations"])
+
+    def test_malformed_json_is_rejected(self):
+        with self.assertRaises(json.JSONDecodeError):
+            parse_model_json("{bad json")
+
+    def test_all_expected_classifications_evaluate_as_pass(self):
+        for case in self.cases:
+            evaluation = evaluate_case(case, classification_for(case))
+            with self.subTest(case=case["case_id"]):
+                self.assertEqual(evaluation["classification"], "PASS")
+
+    def test_fixed_prompt_contains_candidate_and_no_gold_expectation(self):
+        prompt = build_prompt(self.by_id["NA-03"])
+        self.assertIn("candidate observation is obs_2", prompt)
+        self.assertNotIn("expected", prompt)
+        self.assertNotIn("Who is responsible", prompt)
+
+    def test_fixed_model_configuration(self):
+        payload = build_ollama_payload("qwen3.5:9B", "prompt", 16384, 1024)
+        self.assertFalse(payload["think"])
+        self.assertFalse(payload["stream"])
+        self.assertEqual(payload["options"]["temperature"], 0)
+
+    def test_no_rejection_or_status_derivation_function_exists(self):
+        public_names = {name for name in dir(module) if not name.startswith("_")}
+        self.assertNotIn("derive_rejection", public_names)
+        self.assertFalse(any(name.startswith("derive_") for name in public_names))
+
+    def test_artifacts_preserve_semantic_classification_only(self):
+        case = self.by_id["NA-01"]
+        raw = json.dumps(classification_for(case), ensure_ascii=False)
+        with tempfile.TemporaryDirectory() as temporary:
+            output = Path(temporary) / "run"
+            args = argparse.Namespace(
+                cases=GOLD_PATH, output=output, model="qwen3.5:9B",
+                endpoint="http://unused", timeout=1, num_ctx=16384, num_predict=1024,
+            )
+            with patch.object(module, "load_gold_cases", return_value=[deepcopy(case)]), patch.object(
+                module, "call_ollama", return_value=(raw, {"model": "qwen3.5:9B"})
+            ):
+                summary = run_experiment(args)
+            self.assertEqual(summary["successful_llm_call_count"], 1)
+            case_dir = output / "na-01"
+            for filename in (
+                "v3_style_input_observations.json", "prompt.txt", "raw_model_response.txt",
+                "parsed_semantic_classification.json", "structural_validation.json",
+                "evaluation.json", "ollama_metadata.json",
+            ):
+                self.assertTrue((case_dir / filename).is_file(), filename)
+            self.assertFalse((case_dir / "final_derived_result.json").exists())
+            self.assertFalse((case_dir / "deterministic_gate_results.json").exists())
+            parsed = json.loads((case_dir / "parsed_semantic_classification.json").read_text())
+            self.assertEqual(set(parsed), {"observation_id", "negative_act_form", "normalized_action_text"})
+
+
+if __name__ == "__main__":
+    unittest.main()