Remove confidence scoring and refine correction prompts

This commit is contained in:
2026-11-11 17:05:52 +00:00
parent bbf89bb4f5
commit d155bf4cb3
92 changed files with 448 additions and 786 deletions

View File

@@ -74,7 +74,6 @@ func TestGlossaryModuleValidatorChain(t *testing.T) {
"proposal_shape",
"no_effect",
"original_text_presence",
"confidence_threshold",
"protected_terms",
"non_empty_resulting_segment",
"spoken_form_plausibility",
@@ -91,7 +90,7 @@ func TestGlossaryModuleProposeMapsCorrectionsAndWritesDiagnostics(t *testing.T)
responses: []proposal_generation.StructuredCorrectionSet{
{
Corrections: []proposal_generation.StructuredCorrectionProposal{
{SegmentReference: 1, OriginalText: "gestures", CorrectedText: "Jesters", Confidence: 0.94},
{SegmentReference: 1, OriginalText: "gestures", CorrectedText: "Jesters"},
},
},
},

View File

@@ -44,7 +44,6 @@ func (f *fakeLLMClient) CompleteStructured(ctx context.Context, req contracts.St
SegmentReference: item["id"].(int),
OriginalText: item["original_text"].(string),
CorrectedText: item["corrected_text"].(string),
Confidence: item["confidence"].(float64),
})
}
return contracts.StructuredCompletionResponse{}, nil
@@ -84,7 +83,6 @@ func TestGrammarModuleValidatorChain(t *testing.T) {
"proposal_shape",
"no_effect",
"original_text_presence",
"confidence_threshold",
"protected_terms",
"non_empty_resulting_segment",
"editorial_review",
@@ -101,7 +99,7 @@ func TestGrammarModuleProposeMapsCorrectionsAndWritesDiagnostics(t *testing.T) {
responses: []map[string]any{
{
"corrections": []map[string]any{
{"id": 1, "original_text": "hello ,world", "corrected_text": "Hello, world", "confidence": 0.93},
{"id": 1, "original_text": "hello ,world", "corrected_text": "Hello, world"},
},
},
},

View File

@@ -8,7 +8,6 @@ import (
"strings"
"testing"
"gitea.maximumdirect.net/eric/audita/v3/internal/core/config"
domain "gitea.maximumdirect.net/eric/audita/v3/internal/core/glossary"
"gitea.maximumdirect.net/eric/audita/v3/internal/core/schema"
"gitea.maximumdirect.net/eric/audita/v3/internal/framework/contracts"
@@ -76,7 +75,6 @@ func TestHomophonesModuleValidatorChain(t *testing.T) {
"proposal_shape",
"no_effect",
"original_text_presence",
"confidence_threshold",
"protected_terms",
"non_empty_resulting_segment",
"spoken_form_plausibility",
@@ -93,7 +91,7 @@ func TestHomophonesModuleProposeMapsCorrectionsAndWritesDiagnostics(t *testing.T
responses: []proposal_generation.StructuredCorrectionSet{
{
Corrections: []proposal_generation.StructuredCorrectionProposal{
{SegmentReference: 1, OriginalText: "gestures", CorrectedText: "Jesters", Confidence: 0.94},
{SegmentReference: 1, OriginalText: "gestures", CorrectedText: "Jesters"},
},
},
},
@@ -150,51 +148,6 @@ func TestHomophonesModuleProposeMapsCorrectionsAndWritesDiagnostics(t *testing.T
}
}
func TestHomophonesConfidenceThresholdIsUsed(t *testing.T) {
module, err := New()
if err != nil {
t.Fatalf("New error: %v", err)
}
var thresholdValidator contracts.Validator
for _, v := range module.Validators() {
if v.Name() == "confidence_threshold" {
thresholdValidator = v
break
}
}
if thresholdValidator == nil {
t.Fatal("expected confidence_threshold validator")
}
result, err := thresholdValidator.Validate(context.Background(), validators.Request{
Settings: config.Stage{ConfidenceThreshold: 0.95},
ModuleKey: "homophones",
CandidateProposal: []proposals.EnrichedCorrectionProposal{
{
CorrectionProposal: proposals.CorrectionProposal{
TargetSegmentID: 1,
OriginalText: "gestures",
CorrectedText: "Jesters",
Confidence: 0.94,
},
ProposalMetadata: proposals.ProposalMetadata{ProposalIndex: 0, ModuleKey: "homophones", ModuleInstance: "homophones"},
},
},
})
if err != nil {
t.Fatalf("validate error: %v", err)
}
if len(result.Decisions) != 1 {
t.Fatalf("expected one decision, got %+v", result.Decisions)
}
if result.Decisions[0].Approved {
t.Fatalf("expected rejection under homophones threshold, got %+v", result.Decisions[0])
}
if result.Decisions[0].ReasonCode != validators.ReasonLowConfidence {
t.Fatalf("expected low confidence reason, got %+v", result.Decisions[0])
}
}
func TestHomophonesProtectedGlossaryTermBehaviorRejectsUnsafeCorrection(t *testing.T) {
module, err := New()
if err != nil {
@@ -226,7 +179,6 @@ func TestHomophonesProtectedGlossaryTermBehaviorRejectsUnsafeCorrection(t *testi
TargetSegmentID: 1,
OriginalText: "Jesters",
CorrectedText: "gestures",
Confidence: 0.99,
},
ProposalMetadata: proposals.ProposalMetadata{ProposalIndex: 0, ModuleKey: "homophones", ModuleInstance: "homophones"},
},

View File

@@ -8,7 +8,6 @@ import (
"strings"
"testing"
"gitea.maximumdirect.net/eric/audita/v3/internal/core/config"
domain "gitea.maximumdirect.net/eric/audita/v3/internal/core/glossary"
"gitea.maximumdirect.net/eric/audita/v3/internal/core/schema"
"gitea.maximumdirect.net/eric/audita/v3/internal/framework/contracts"
@@ -76,7 +75,6 @@ func TestSpokenWordModuleValidatorChain(t *testing.T) {
"proposal_shape",
"no_effect",
"original_text_presence",
"confidence_threshold",
"protected_terms",
"non_empty_resulting_segment",
"editorial_review",
@@ -93,7 +91,7 @@ func TestSpokenWordModuleProposeMapsCorrectionsAndWritesDiagnostics(t *testing.T
responses: []proposal_generation.StructuredCorrectionSet{
{
Corrections: []proposal_generation.StructuredCorrectionProposal{
{SegmentReference: 1, OriginalText: "I I think", CorrectedText: "I think", Confidence: 0.94},
{SegmentReference: 1, OriginalText: "I I think", CorrectedText: "I think"},
},
},
},
@@ -150,51 +148,6 @@ func TestSpokenWordModuleProposeMapsCorrectionsAndWritesDiagnostics(t *testing.T
}
}
func TestSpokenWordConfidenceThresholdIsUsed(t *testing.T) {
module, err := New()
if err != nil {
t.Fatalf("New error: %v", err)
}
var thresholdValidator contracts.Validator
for _, v := range module.Validators() {
if v.Name() == "confidence_threshold" {
thresholdValidator = v
break
}
}
if thresholdValidator == nil {
t.Fatal("expected confidence_threshold validator")
}
result, err := thresholdValidator.Validate(context.Background(), validators.Request{
Settings: config.Stage{ConfidenceThreshold: 0.95},
ModuleKey: "spoken_word",
CandidateProposal: []proposals.EnrichedCorrectionProposal{
{
CorrectionProposal: proposals.CorrectionProposal{
TargetSegmentID: 1,
OriginalText: "I I think",
CorrectedText: "I think",
Confidence: 0.94,
},
ProposalMetadata: proposals.ProposalMetadata{ProposalIndex: 0, ModuleKey: "spoken_word", ModuleInstance: "spoken_word"},
},
},
})
if err != nil {
t.Fatalf("validate error: %v", err)
}
if len(result.Decisions) != 1 {
t.Fatalf("expected one decision, got %+v", result.Decisions)
}
if result.Decisions[0].Approved {
t.Fatalf("expected rejection under spoken_word threshold, got %+v", result.Decisions[0])
}
if result.Decisions[0].ReasonCode != validators.ReasonLowConfidence {
t.Fatalf("expected low confidence reason, got %+v", result.Decisions[0])
}
}
func TestSpokenWordProtectedGlossaryTermBehaviorRejectsUnsafeCorrection(t *testing.T) {
module, err := New()
if err != nil {
@@ -226,7 +179,6 @@ func TestSpokenWordProtectedGlossaryTermBehaviorRejectsUnsafeCorrection(t *testi
TargetSegmentID: 1,
OriginalText: "Jesters",
CorrectedText: "gestures",
Confidence: 0.99,
},
ProposalMetadata: proposals.ProposalMetadata{ProposalIndex: 0, ModuleKey: "spoken_word", ModuleInstance: "spoken_word"},
},