causantic 0.2.0

package/dist/eval/experiments/topic-continuity/hybrid-classifier.js

@@ -0,0 +1,175 @@
+/**
+ * Hybrid classifier combining embedding distance with lexical features.
+ *
+ * Uses tunable weights to combine multiple signals for topic continuity
+ * prediction.
+ */
+import { extractLexicalFeatures, computeLexicalScore } from './lexical-features.js';
+export const DEFAULT_WEIGHTS = {
+    embedding: 0.5,
+    topicShiftMarker: 0.2,
+    continuationMarker: 0.15,
+    timeGap: 0.05,
+    filePathOverlap: 0.05,
+    keywordOverlap: 0.05,
+};
+/**
+ * Compute hybrid continuation score from features.
+ */
+export function computeHybridScore(features, weights = DEFAULT_WEIGHTS) {
+    let score = 0;
+    let totalWeight = 0;
+    // Embedding distance (inverted: low distance = high continuation score)
+    const embeddingScore = 1 - features.embeddingDistanceMin;
+    score += weights.embedding * embeddingScore;
+    totalWeight += weights.embedding;
+    // Topic shift marker (binary: presence indicates new topic)
+    const topicShiftScore = features.hasTopicShiftMarker ? 0 : 1;
+    score += weights.topicShiftMarker * topicShiftScore;
+    totalWeight += weights.topicShiftMarker;
+    // Continuation marker (binary: presence indicates continuation)
+    const continuationScore = features.hasContinuationMarker ? 1 : 0.5;
+    score += weights.continuationMarker * continuationScore;
+    totalWeight += weights.continuationMarker;
+    // Time gap (sigmoid decay: longer gap = lower continuation score)
+    const timeGapScore = timeGapToScore(features.timeGapMinutes);
+    score += weights.timeGap * timeGapScore;
+    totalWeight += weights.timeGap;
+    // File path overlap (direct: more overlap = higher continuation)
+    score += weights.filePathOverlap * features.filePathOverlap;
+    totalWeight += weights.filePathOverlap;
+    // Keyword overlap (direct: more overlap = higher continuation)
+    score += weights.keywordOverlap * features.keywordOverlap;
+    totalWeight += weights.keywordOverlap;
+    // Normalize by total weight
+    return totalWeight > 0 ? score / totalWeight : 0.5;
+}
+/**
+ * Convert time gap to a continuation score using sigmoid decay.
+ * Short gaps -> high score, long gaps -> low score.
+ */
+function timeGapToScore(timeGapMinutes) {
+    // Sigmoid centered at 30 minutes
+    const k = 0.1; // Steepness
+    const midpoint = 30;
+    return 1 / (1 + Math.exp(k * (timeGapMinutes - midpoint)));
+}
+/**
+ * Classify transitions using hybrid approach (requires pre-embedded data).
+ */
+export function classifyWithHybrid(embeddedTransitions, weights = DEFAULT_WEIGHTS) {
+    return embeddedTransitions.map((et) => {
+        const lexicalFeatures = extractLexicalFeatures(et.transition.prevAssistantText, et.transition.nextUserText, et.transition.timeGapMs);
+        const features = {
+            ...lexicalFeatures,
+            embeddingDistanceMin: et.embeddingDistanceMin,
+            embeddingDistanceMean: et.embeddingDistanceMean,
+        };
+        const score = computeHybridScore(features, weights);
+        return {
+            transitionId: et.transition.id,
+            groundTruth: et.transition.label,
+            continuationScore: score,
+            features,
+        };
+    });
+}
+/**
+ * Classify using lexical features only (no embedding).
+ */
+export function classifyWithLexicalOnly(transitions) {
+    return transitions
+        .filter((t) => t.prevAssistantText.trim()) // Skip transitions without prior context
+        .map((t) => {
+        const lexicalFeatures = extractLexicalFeatures(t.prevAssistantText, t.nextUserText, t.timeGapMs);
+        const features = {
+            ...lexicalFeatures,
+            embeddingDistanceMin: 0.5, // Neutral placeholder
+            embeddingDistanceMean: 0.5,
+        };
+        const score = computeLexicalScore(lexicalFeatures);
+        return {
+            transitionId: t.id,
+            groundTruth: t.label,
+            continuationScore: score,
+            features,
+        };
+    });
+}
+export const ALL_FEATURES = {
+    useEmbedding: true,
+    useTopicShiftMarker: true,
+    useContinuationMarker: true,
+    useTimeGap: true,
+    useFilePathOverlap: true,
+    useKeywordOverlap: true,
+};
+export const EMBEDDING_ONLY = {
+    useEmbedding: true,
+    useTopicShiftMarker: false,
+    useContinuationMarker: false,
+    useTimeGap: false,
+    useFilePathOverlap: false,
+    useKeywordOverlap: false,
+};
+export const LEXICAL_ONLY = {
+    useEmbedding: false,
+    useTopicShiftMarker: true,
+    useContinuationMarker: true,
+    useTimeGap: true,
+    useFilePathOverlap: true,
+    useKeywordOverlap: true,
+};
+/**
+ * Compute score with specific features enabled/disabled.
+ */
+export function computeAblatedScore(features, flags, baseWeights = DEFAULT_WEIGHTS) {
+    // Zero out disabled features
+    const weights = {
+        embedding: flags.useEmbedding ? baseWeights.embedding : 0,
+        topicShiftMarker: flags.useTopicShiftMarker ? baseWeights.topicShiftMarker : 0,
+        continuationMarker: flags.useContinuationMarker ? baseWeights.continuationMarker : 0,
+        timeGap: flags.useTimeGap ? baseWeights.timeGap : 0,
+        filePathOverlap: flags.useFilePathOverlap ? baseWeights.filePathOverlap : 0,
+        keywordOverlap: flags.useKeywordOverlap ? baseWeights.keywordOverlap : 0,
+    };
+    return computeHybridScore(features, weights);
+}
+/**
+ * Classify with specific features enabled for ablation study.
+ */
+export function classifyWithAblation(embeddedTransitions, flags) {
+    return embeddedTransitions.map((et) => {
+        const lexicalFeatures = extractLexicalFeatures(et.transition.prevAssistantText, et.transition.nextUserText, et.transition.timeGapMs);
+        const features = {
+            ...lexicalFeatures,
+            embeddingDistanceMin: et.embeddingDistanceMin,
+            embeddingDistanceMean: et.embeddingDistanceMean,
+        };
+        const score = computeAblatedScore(features, flags);
+        return {
+            transitionId: et.transition.id,
+            groundTruth: et.transition.label,
+            continuationScore: score,
+            features,
+        };
+    });
+}
+/**
+ * Named ablation configurations for experiment.
+ */
+export const ABLATION_CONFIGS = [
+    { name: 'time-gap-only', flags: { ...LEXICAL_ONLY, useTopicShiftMarker: false, useContinuationMarker: false, useFilePathOverlap: false, useKeywordOverlap: false } },
+    { name: 'shift-markers-only', flags: { ...LEXICAL_ONLY, useTimeGap: false, useContinuationMarker: false, useFilePathOverlap: false, useKeywordOverlap: false } },
+    { name: 'continuation-markers-only', flags: { ...LEXICAL_ONLY, useTimeGap: false, useTopicShiftMarker: false, useFilePathOverlap: false, useKeywordOverlap: false } },
+    { name: 'file-path-overlap-only', flags: { ...LEXICAL_ONLY, useTimeGap: false, useTopicShiftMarker: false, useContinuationMarker: false, useKeywordOverlap: false } },
+    { name: 'keyword-overlap-only', flags: { ...LEXICAL_ONLY, useTimeGap: false, useTopicShiftMarker: false, useContinuationMarker: false, useFilePathOverlap: false } },
+    { name: 'all-lexical', flags: LEXICAL_ONLY },
+    { name: 'embedding-only', flags: EMBEDDING_ONLY },
+    { name: 'embedding+time-gap', flags: { ...EMBEDDING_ONLY, useTimeGap: true } },
+    { name: 'embedding+markers', flags: { ...EMBEDDING_ONLY, useTopicShiftMarker: true, useContinuationMarker: true } },
+    { name: 'embedding+paths', flags: { ...EMBEDDING_ONLY, useFilePathOverlap: true } },
+    { name: 'embedding+keywords', flags: { ...EMBEDDING_ONLY, useKeywordOverlap: true } },
+    { name: 'all-features', flags: ALL_FEATURES },
+];
+//# sourceMappingURL=hybrid-classifier.js.map

package/dist/eval/experiments/topic-continuity/hybrid-classifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"hybrid-classifier.js","sourceRoot":"","sources":["../../../../src/eval/experiments/topic-continuity/hybrid-classifier.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAOH,OAAO,EAAE,sBAAsB,EAAE,mBAAmB,EAAE,MAAM,uBAAuB,CAAC;AAwBpF,MAAM,CAAC,MAAM,eAAe,GAAkB;IAC5C,SAAS,EAAE,GAAG;IACd,gBAAgB,EAAE,GAAG;IACrB,kBAAkB,EAAE,IAAI;IACxB,OAAO,EAAE,IAAI;IACb,eAAe,EAAE,IAAI;IACrB,cAAc,EAAE,IAAI;CACrB,CAAC;AAEF;;GAEG;AACH,MAAM,UAAU,kBAAkB,CAChC,QAA4B,EAC5B,UAAyB,eAAe;IAExC,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,IAAI,WAAW,GAAG,CAAC,CAAC;IAEpB,wEAAwE;IACxE,MAAM,cAAc,GAAG,CAAC,GAAG,QAAQ,CAAC,oBAAoB,CAAC;IACzD,KAAK,IAAI,OAAO,CAAC,SAAS,GAAG,cAAc,CAAC;IAC5C,WAAW,IAAI,OAAO,CAAC,SAAS,CAAC;IAEjC,4DAA4D;IAC5D,MAAM,eAAe,GAAG,QAAQ,CAAC,mBAAmB,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC7D,KAAK,IAAI,OAAO,CAAC,gBAAgB,GAAG,eAAe,CAAC;IACpD,WAAW,IAAI,OAAO,CAAC,gBAAgB,CAAC;IAExC,gEAAgE;IAChE,MAAM,iBAAiB,GAAG,QAAQ,CAAC,qBAAqB,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC;IACnE,KAAK,IAAI,OAAO,CAAC,kBAAkB,GAAG,iBAAiB,CAAC;IACxD,WAAW,IAAI,OAAO,CAAC,kBAAkB,CAAC;IAE1C,kEAAkE;IAClE,MAAM,YAAY,GAAG,cAAc,CAAC,QAAQ,CAAC,cAAc,CAAC,CAAC;IAC7D,KAAK,IAAI,OAAO,CAAC,OAAO,GAAG,YAAY,CAAC;IACxC,WAAW,IAAI,OAAO,CAAC,OAAO,CAAC;IAE/B,iEAAiE;IACjE,KAAK,IAAI,OAAO,CAAC,eAAe,GAAG,QAAQ,CAAC,eAAe,CAAC;IAC5D,WAAW,IAAI,OAAO,CAAC,eAAe,CAAC;IAEvC,+DAA+D;IAC/D,KAAK,IAAI,OAAO,CAAC,cAAc,GAAG,QAAQ,CAAC,cAAc,CAAC;IAC1D,WAAW,IAAI,OAAO,CAAC,cAAc,CAAC;IAEtC,4BAA4B;IAC5B,OAAO,WAAW,GAAG,CAAC,CAAC,CAAC,CAAC,KAAK,GAAG,WAAW,CAAC,CAAC,CAAC,GAAG,CAAC;AACrD,CAAC;AAED;;;GAGG;AACH,SAAS,cAAc,CAAC,cAAsB;IAC5C,iCAAiC;IACjC,MAAM,CAAC,GAAG,GAAG,CAAC,CAAC,YAAY;IAC3B,MAAM,QAAQ,GAAG,EAAE,CAAC;IACpB,OAAO,CAAC,GAAG,CAAC,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,cAAc,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;AAC7D,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,kBAAkB,CAChC,mBAAyC,EACzC,UAAyB,eAAe;IAExC,OAAO,mBAAmB,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE;QACpC,MAAM,eAAe,GAAG,sBAAsB,CAC5C,EAAE,CAAC,UAAU,CAAC,iBAAiB,EAC/B,EAAE,CAAC,UAAU,CAAC,YAAY,EAC1B,EAAE,CAAC,UAAU,CAAC,SAAS,CACxB,CAAC;QAEF,MAAM,QAAQ,GAAuB;YACnC,GAAG,eAAe;YAClB,oBAAoB,EAAE,EAAE,CAAC,oBAAoB;YAC7C,qBAAqB,EAAE,EAAE,CAAC,qBAAqB;SAChD,CAAC;QAEF,MAAM,KAAK,GAAG,kBAAkB,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;QAEpD,OAAO;YACL,YAAY,EAAE,EAAE,CAAC,UAAU,CAAC,EAAE;YAC9B,WAAW,EAAE,EAAE,CAAC,UAAU,CAAC,KAAK;YAChC,iBAAiB,EAAE,KAAK;YACxB,QAAQ;SACT,CAAC;IACJ,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,uBAAuB,CACrC,WAA6B;IAE7B,OAAO,WAAW;SACf,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,iBAAiB,CAAC,IAAI,EAAE,CAAC,CAAC,yCAAyC;SACnF,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE;QACT,MAAM,eAAe,GAAG,sBAAsB,CAC5C,CAAC,CAAC,iBAAiB,EACnB,CAAC,CAAC,YAAY,EACd,CAAC,CAAC,SAAS,CACZ,CAAC;QAEF,MAAM,QAAQ,GAAuB;YACnC,GAAG,eAAe;YAClB,oBAAoB,EAAE,GAAG,EAAE,sBAAsB;YACjD,qBAAqB,EAAE,GAAG;SAC3B,CAAC;QAEF,MAAM,KAAK,GAAG,mBAAmB,CAAC,eAAe,CAAC,CAAC;QAEnD,OAAO;YACL,YAAY,EAAE,CAAC,CAAC,EAAE;YAClB,WAAW,EAAE,CAAC,CAAC,KAAK;YACpB,iBAAiB,EAAE,KAAK;YACxB,QAAQ;SACT,CAAC;IACJ,CAAC,CAAC,CAAC;AACP,CAAC;AAcD,MAAM,CAAC,MAAM,YAAY,GAAiB;IACxC,YAAY,EAAE,IAAI;IAClB,mBAAmB,EAAE,IAAI;IACzB,qBAAqB,EAAE,IAAI;IAC3B,UAAU,EAAE,IAAI;IAChB,kBAAkB,EAAE,IAAI;IACxB,iBAAiB,EAAE,IAAI;CACxB,CAAC;AAEF,MAAM,CAAC,MAAM,cAAc,GAAiB;IAC1C,YAAY,EAAE,IAAI;IAClB,mBAAmB,EAAE,KAAK;IAC1B,qBAAqB,EAAE,KAAK;IAC5B,UAAU,EAAE,KAAK;IACjB,kBAAkB,EAAE,KAAK;IACzB,iBAAiB,EAAE,KAAK;CACzB,CAAC;AAEF,MAAM,CAAC,MAAM,YAAY,GAAiB;IACxC,YAAY,EAAE,KAAK;IACnB,mBAAmB,EAAE,IAAI;IACzB,qBAAqB,EAAE,IAAI;IAC3B,UAAU,EAAE,IAAI;IAChB,kBAAkB,EAAE,IAAI;IACxB,iBAAiB,EAAE,IAAI;CACxB,CAAC;AAEF;;GAEG;AACH,MAAM,UAAU,mBAAmB,CACjC,QAA4B,EAC5B,KAAmB,EACnB,cAA6B,eAAe;IAE5C,6BAA6B;IAC7B,MAAM,OAAO,GAAkB;QAC7B,SAAS,EAAE,KAAK,CAAC,YAAY,CAAC,CAAC,CAAC,WAAW,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC;QACzD,gBAAgB,EAAE,KAAK,CAAC,mBAAmB,CAAC,CAAC,CAAC,WAAW,CAAC,gBAAgB,CAAC,CAAC,CAAC,CAAC;QAC9E,kBAAkB,EAAE,KAAK,CAAC,qBAAqB,CAAC,CAAC,CAAC,WAAW,CAAC,kBAAkB,CAAC,CAAC,CAAC,CAAC;QACpF,OAAO,EAAE,KAAK,CAAC,UAAU,CAAC,CAAC,CAAC,WAAW,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;QACnD,eAAe,EAAE,KAAK,CAAC,kBAAkB,CAAC,CAAC,CAAC,WAAW,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC;QAC3E,cAAc,EAAE,KAAK,CAAC,iBAAiB,CAAC,CAAC,CAAC,WAAW,CAAC,cAAc,CAAC,CAAC,CAAC,CAAC;KACzE,CAAC;IAEF,OAAO,kBAAkB,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;AAC/C,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,oBAAoB,CAClC,mBAAyC,EACzC,KAAmB;IAEnB,OAAO,mBAAmB,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE;QACpC,MAAM,eAAe,GAAG,sBAAsB,CAC5C,EAAE,CAAC,UAAU,CAAC,iBAAiB,EAC/B,EAAE,CAAC,UAAU,CAAC,YAAY,EAC1B,EAAE,CAAC,UAAU,CAAC,SAAS,CACxB,CAAC;QAEF,MAAM,QAAQ,GAAuB;YACnC,GAAG,eAAe;YAClB,oBAAoB,EAAE,EAAE,CAAC,oBAAoB;YAC7C,qBAAqB,EAAE,EAAE,CAAC,qBAAqB;SAChD,CAAC;QAEF,MAAM,KAAK,GAAG,mBAAmB,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC;QAEnD,OAAO;YACL,YAAY,EAAE,EAAE,CAAC,UAAU,CAAC,EAAE;YAC9B,WAAW,EAAE,EAAE,CAAC,UAAU,CAAC,KAAK;YAChC,iBAAiB,EAAE,KAAK;YACxB,QAAQ;SACT,CAAC;IACJ,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAA4C;IACvE,EAAE,IAAI,EAAE,eAAe,EAAE,KAAK,EAAE,EAAE,GAAG,YAAY,EAAE,mBAAmB,EAAE,KAAK,EAAE,qBAAqB,EAAE,KAAK,EAAE,kBAAkB,EAAE,KAAK,EAAE,iBAAiB,EAAE,KAAK,EAAE,EAAE;IACpK,EAAE,IAAI,EAAE,oBAAoB,EAAE,KAAK,EAAE,EAAE,GAAG,YAAY,EAAE,UAAU,EAAE,KAAK,EAAE,qBAAqB,EAAE,KAAK,EAAE,kBAAkB,EAAE,KAAK,EAAE,iBAAiB,EAAE,KAAK,EAAE,EAAE;IAChK,EAAE,IAAI,EAAE,2BAA2B,EAAE,KAAK,EAAE,EAAE,GAAG,YAAY,EAAE,UAAU,EAAE,KAAK,EAAE,mBAAmB,EAAE,KAAK,EAAE,kBAAkB,EAAE,KAAK,EAAE,iBAAiB,EAAE,KAAK,EAAE,EAAE;IACrK,EAAE,IAAI,EAAE,wBAAwB,EAAE,KAAK,EAAE,EAAE,GAAG,YAAY,EAAE,UAAU,EAAE,KAAK,EAAE,mBAAmB,EAAE,KAAK,EAAE,qBAAqB,EAAE,KAAK,EAAE,iBAAiB,EAAE,KAAK,EAAE,EAAE;IACrK,EAAE,IAAI,EAAE,sBAAsB,EAAE,KAAK,EAAE,EAAE,GAAG,YAAY,EAAE,UAAU,EAAE,KAAK,EAAE,mBAAmB,EAAE,KAAK,EAAE,qBAAqB,EAAE,KAAK,EAAE,kBAAkB,EAAE,KAAK,EAAE,EAAE;IACpK,EAAE,IAAI,EAAE,aAAa,EAAE,KAAK,EAAE,YAAY,EAAE;IAC5C,EAAE,IAAI,EAAE,gBAAgB,EAAE,KAAK,EAAE,cAAc,EAAE;IACjD,EAAE,IAAI,EAAE,oBAAoB,EAAE,KAAK,EAAE,EAAE,GAAG,cAAc,EAAE,UAAU,EAAE,IAAI,EAAE,EAAE;IAC9E,EAAE,IAAI,EAAE,mBAAmB,EAAE,KAAK,EAAE,EAAE,GAAG,cAAc,EAAE,mBAAmB,EAAE,IAAI,EAAE,qBAAqB,EAAE,IAAI,EAAE,EAAE;IACnH,EAAE,IAAI,EAAE,iBAAiB,EAAE,KAAK,EAAE,EAAE,GAAG,cAAc,EAAE,kBAAkB,EAAE,IAAI,EAAE,EAAE;IACnF,EAAE,IAAI,EAAE,oBAAoB,EAAE,KAAK,EAAE,EAAE,GAAG,cAAc,EAAE,iBAAiB,EAAE,IAAI,EAAE,EAAE;IACrF,EAAE,IAAI,EAAE,cAAc,EAAE,KAAK,EAAE,YAAY,EAAE;CAC9C,CAAC"}

package/dist/eval/experiments/topic-continuity/index.d.ts

@@ -0,0 +1,13 @@
+/**
+ * Topic continuity detection experiment module.
+ *
+ * Classifies whether a user's message continues the previous topic
+ * or starts a new topic, supporting the D-T-D model's edge creation logic.
+ */
+export type { TransitionLabel, Confidence, TurnTransition, TransitionFeatures, ClassificationResult, ClassifierMetrics, ModelResults, FeatureAblation, DatasetStats, ExperimentReport, ThresholdSweepRow, TimeGapSweepRow, } from './types.js';
+export { generateTransitionLabels, generateSessionTransitions, computeDatasetStats, filterHighConfidence, balanceDataset, type SessionSource, type LabelerOptions, } from './labeler.js';
+export { hasTopicShiftMarker, hasContinuationMarker, extractFilePaths, computeFilePathOverlap, extractKeywords, computeKeywordOverlap, extractLexicalFeatures, computeLexicalScore, classifyWithLexicalFeatures, type LexicalClassificationResult, } from './lexical-features.js';
+export { embedTransitions, classifyWithEmbeddings, embeddingDistanceToScore, getEmbeddingDistances, createEmbeddingCache, classifyFromCache, type EmbeddedTransition, type EmbeddingCache, } from './embedding-classifier.js';
+export { computeHybridScore, classifyWithHybrid, classifyWithLexicalOnly, classifyWithAblation, DEFAULT_WEIGHTS, ALL_FEATURES, EMBEDDING_ONLY, LEXICAL_ONLY, ABLATION_CONFIGS, type HybridWeights, type FeatureFlags, } from './hybrid-classifier.js';
+export { runTopicContinuityExperiment, computeRocAuc, computeMetrics, findOptimalThreshold, exportTransitionsDataset, type ExperimentOptions, } from './run-experiment.js';
+//# sourceMappingURL=index.d.ts.map

package/dist/eval/experiments/topic-continuity/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../../src/eval/experiments/topic-continuity/index.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAGH,YAAY,EACV,eAAe,EACf,UAAU,EACV,cAAc,EACd,kBAAkB,EAClB,oBAAoB,EACpB,iBAAiB,EACjB,YAAY,EACZ,eAAe,EACf,YAAY,EACZ,gBAAgB,EAChB,iBAAiB,EACjB,eAAe,GAChB,MAAM,YAAY,CAAC;AAGpB,OAAO,EACL,wBAAwB,EACxB,0BAA0B,EAC1B,mBAAmB,EACnB,oBAAoB,EACpB,cAAc,EACd,KAAK,aAAa,EAClB,KAAK,cAAc,GACpB,MAAM,cAAc,CAAC;AAGtB,OAAO,EACL,mBAAmB,EACnB,qBAAqB,EACrB,gBAAgB,EAChB,sBAAsB,EACtB,eAAe,EACf,qBAAqB,EACrB,sBAAsB,EACtB,mBAAmB,EACnB,2BAA2B,EAC3B,KAAK,2BAA2B,GACjC,MAAM,uBAAuB,CAAC;AAG/B,OAAO,EACL,gBAAgB,EAChB,sBAAsB,EACtB,wBAAwB,EACxB,qBAAqB,EACrB,oBAAoB,EACpB,iBAAiB,EACjB,KAAK,kBAAkB,EACvB,KAAK,cAAc,GACpB,MAAM,2BAA2B,CAAC;AAGnC,OAAO,EACL,kBAAkB,EAClB,kBAAkB,EAClB,uBAAuB,EACvB,oBAAoB,EACpB,eAAe,EACf,YAAY,EACZ,cAAc,EACd,YAAY,EACZ,gBAAgB,EAChB,KAAK,aAAa,EAClB,KAAK,YAAY,GAClB,MAAM,wBAAwB,CAAC;AAGhC,OAAO,EACL,4BAA4B,EAC5B,aAAa,EACb,cAAc,EACd,oBAAoB,EACpB,wBAAwB,EACxB,KAAK,iBAAiB,GACvB,MAAM,qBAAqB,CAAC"}

package/dist/eval/experiments/topic-continuity/index.js

@@ -0,0 +1,17 @@
+/**
+ * Topic continuity detection experiment module.
+ *
+ * Classifies whether a user's message continues the previous topic
+ * or starts a new topic, supporting the D-T-D model's edge creation logic.
+ */
+// Labeling
+export { generateTransitionLabels, generateSessionTransitions, computeDatasetStats, filterHighConfidence, balanceDataset, } from './labeler.js';
+// Lexical features
+export { hasTopicShiftMarker, hasContinuationMarker, extractFilePaths, computeFilePathOverlap, extractKeywords, computeKeywordOverlap, extractLexicalFeatures, computeLexicalScore, classifyWithLexicalFeatures, } from './lexical-features.js';
+// Embedding classifier
+export { embedTransitions, classifyWithEmbeddings, embeddingDistanceToScore, getEmbeddingDistances, createEmbeddingCache, classifyFromCache, } from './embedding-classifier.js';
+// Hybrid classifier
+export { computeHybridScore, classifyWithHybrid, classifyWithLexicalOnly, classifyWithAblation, DEFAULT_WEIGHTS, ALL_FEATURES, EMBEDDING_ONLY, LEXICAL_ONLY, ABLATION_CONFIGS, } from './hybrid-classifier.js';
+// Experiment runner
+export { runTopicContinuityExperiment, computeRocAuc, computeMetrics, findOptimalThreshold, exportTransitionsDataset, } from './run-experiment.js';
+//# sourceMappingURL=index.js.map

package/dist/eval/experiments/topic-continuity/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../../src/eval/experiments/topic-continuity/index.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAkBH,WAAW;AACX,OAAO,EACL,wBAAwB,EACxB,0BAA0B,EAC1B,mBAAmB,EACnB,oBAAoB,EACpB,cAAc,GAGf,MAAM,cAAc,CAAC;AAEtB,mBAAmB;AACnB,OAAO,EACL,mBAAmB,EACnB,qBAAqB,EACrB,gBAAgB,EAChB,sBAAsB,EACtB,eAAe,EACf,qBAAqB,EACrB,sBAAsB,EACtB,mBAAmB,EACnB,2BAA2B,GAE5B,MAAM,uBAAuB,CAAC;AAE/B,uBAAuB;AACvB,OAAO,EACL,gBAAgB,EAChB,sBAAsB,EACtB,wBAAwB,EACxB,qBAAqB,EACrB,oBAAoB,EACpB,iBAAiB,GAGlB,MAAM,2BAA2B,CAAC;AAEnC,oBAAoB;AACpB,OAAO,EACL,kBAAkB,EAClB,kBAAkB,EAClB,uBAAuB,EACvB,oBAAoB,EACpB,eAAe,EACf,YAAY,EACZ,cAAc,EACd,YAAY,EACZ,gBAAgB,GAGjB,MAAM,wBAAwB,CAAC;AAEhC,oBAAoB;AACpB,OAAO,EACL,4BAA4B,EAC5B,aAAa,EACb,cAAc,EACd,oBAAoB,EACpB,wBAAwB,GAEzB,MAAM,qBAAqB,CAAC"}

package/dist/eval/experiments/topic-continuity/labeler.d.ts

@@ -0,0 +1,42 @@
+/**
+ * Auto-labeling heuristics for turn transitions.
+ *
+ * Generates labeled TurnTransition records from session data using
+ * structural cues, temporal gaps, and explicit markers.
+ */
+import type { TurnTransition, DatasetStats } from './types.js';
+export interface LabelerOptions {
+    /** Time gap threshold in minutes. Default: 30. */
+    timeGapMinutes?: number;
+    /** Include transitions with low confidence. Default: true. */
+    includeLowConfidence?: boolean;
+}
+/**
+ * Generate turn transitions from a single session file.
+ */
+export declare function generateSessionTransitions(sessionPath: string, sessionId: string, sessionSlug: string, options?: LabelerOptions): Promise<TurnTransition[]>;
+/**
+ * Session info for transition generation.
+ */
+export interface SessionSource {
+    path: string;
+    sessionId: string;
+    sessionSlug: string;
+}
+/**
+ * Generate turn transitions from multiple sessions.
+ */
+export declare function generateTransitionLabels(sessions: SessionSource[], options?: LabelerOptions): Promise<TurnTransition[]>;
+/**
+ * Compute dataset statistics from transitions.
+ */
+export declare function computeDatasetStats(transitions: TurnTransition[]): DatasetStats;
+/**
+ * Filter transitions to only include high-confidence labels.
+ */
+export declare function filterHighConfidence(transitions: TurnTransition[]): TurnTransition[];
+/**
+ * Balance dataset by undersampling the majority class.
+ */
+export declare function balanceDataset(transitions: TurnTransition[], seed?: number): TurnTransition[];
+//# sourceMappingURL=labeler.d.ts.map

package/dist/eval/experiments/topic-continuity/labeler.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"labeler.d.ts","sourceRoot":"","sources":["../../../../src/eval/experiments/topic-continuity/labeler.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAKH,OAAO,KAAK,EACV,cAAc,EAGd,YAAY,EACb,MAAM,YAAY,CAAC;AAoCpB,MAAM,WAAW,cAAc;IAC7B,kDAAkD;IAClD,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,8DAA8D;IAC9D,oBAAoB,CAAC,EAAE,OAAO,CAAC;CAChC;AAoGD;;GAEG;AACH,wBAAsB,0BAA0B,CAC9C,WAAW,EAAE,MAAM,EACnB,SAAS,EAAE,MAAM,EACjB,WAAW,EAAE,MAAM,EACnB,OAAO,GAAE,cAAmB,GAC3B,OAAO,CAAC,cAAc,EAAE,CAAC,CAwD3B;AAED;;GAEG;AACH,MAAM,WAAW,aAAa;IAC5B,IAAI,EAAE,MAAM,CAAC;IACb,SAAS,EAAE,MAAM,CAAC;IAClB,WAAW,EAAE,MAAM,CAAC;CACrB;AAED;;GAEG;AACH,wBAAsB,wBAAwB,CAC5C,QAAQ,EAAE,aAAa,EAAE,EACzB,OAAO,GAAE,cAAmB,GAC3B,OAAO,CAAC,cAAc,EAAE,CAAC,CA8B3B;AAED;;GAEG;AACH,wBAAgB,mBAAmB,CAAC,WAAW,EAAE,cAAc,EAAE,GAAG,YAAY,CA6B/E;AAED;;GAEG;AACH,wBAAgB,oBAAoB,CAAC,WAAW,EAAE,cAAc,EAAE,GAAG,cAAc,EAAE,CAEpF;AAED;;GAEG;AACH,wBAAgB,cAAc,CAC5B,WAAW,EAAE,cAAc,EAAE,EAC7B,IAAI,GAAE,MAAW,GAChB,cAAc,EAAE,CAclB"}

package/dist/eval/experiments/topic-continuity/labeler.js

@@ -0,0 +1,253 @@
+/**
+ * Auto-labeling heuristics for turn transitions.
+ *
+ * Generates labeled TurnTransition records from session data using
+ * structural cues, temporal gaps, and explicit markers.
+ */
+import { readSessionMessages } from '../../../parser/session-reader.js';
+import { assembleTurns } from '../../../parser/turn-assembler.js';
+/** Boilerplate prefix indicating a continued session. */
+const CONTINUATION_BOILERPLATE = 'This session is being continued from a previous conversation';
+/** Patterns indicating explicit topic shift. */
+const TOPIC_SHIFT_PATTERNS = [
+    /^actually,?\s*(let'?s?|can we|could we)/i,
+    /^(new|different|another)\s+(question|topic|issue)/i,
+    /^switching\s+(to|gears)/i,
+    /^(on|about)\s+a\s+(different|separate)\s+(note|topic)/i,
+    /^(ok|okay),?\s*(so|now)\s+/i,
+    /^moving\s+on/i,
+    /^forget\s+(about\s+)?(that|this)/i,
+    /^let'?s\s+(change|switch|move)/i,
+    /^(unrelated|separate)\s*(question|topic|issue)?/i,
+    /^(btw|by the way)/i,
+];
+/** Patterns indicating continuation of the previous topic. */
+const CONTINUATION_PATTERNS = [
+    /^(yes|no|right|correct|exactly)/i,
+    /^(it|that|this)\s+(is|was|shows|works|doesn'?t)/i,
+    /^(the|your)\s+(error|output|result|fix|solution)/i,
+    /^(thanks|thank\s+you)/i,
+    /^(ok|okay|got\s+it|makes\s+sense)/i,
+    /^(but|however|although|though)/i,
+    /^(also|additionally|and\s+also)/i,
+    /^(what\s+about|how\s+about)/i,
+    /^(can you|could you)\s+(also|explain|show)/i,
+    /^(I\s+(see|understand|get\s+it))/i,
+];
+/** Default time gap threshold in minutes for labeling as new topic. */
+const DEFAULT_TIME_GAP_MINUTES = 30;
+/**
+ * Extract plain text from assistant content blocks.
+ */
+function extractAssistantText(blocks) {
+    return blocks
+        .filter((b) => b.type === 'text')
+        .map((b) => b.text)
+        .join('\n');
+}
+/**
+ * Check if text contains continuation boilerplate (resumed session).
+ */
+function hasContinuationBoilerplate(text) {
+    return text.includes(CONTINUATION_BOILERPLATE);
+}
+/**
+ * Check if text contains topic-shift markers.
+ */
+function hasTopicShiftMarker(text) {
+    const trimmed = text.trim();
+    return TOPIC_SHIFT_PATTERNS.some((p) => p.test(trimmed));
+}
+/**
+ * Check if text contains continuation markers.
+ */
+function hasContinuationMarker(text) {
+    const trimmed = text.trim();
+    return CONTINUATION_PATTERNS.some((p) => p.test(trimmed));
+}
+/**
+ * Parse ISO timestamp to milliseconds.
+ */
+function parseTime(iso) {
+    return new Date(iso).getTime();
+}
+/**
+ * Generate a unique transition ID.
+ */
+function makeTransitionId(sessionId, prevIdx, nextIdx) {
+    const shortSession = sessionId.slice(0, 8);
+    return `tr-${shortSession}-${prevIdx}-${nextIdx}`;
+}
+/**
+ * Determine label and confidence for a transition.
+ */
+function labelTransition(prevTurn, nextTurn, isFirstTurn, timeGapMs, timeGapThresholdMinutes) {
+    const userText = nextTurn.userText;
+    const timeGapMinutes = timeGapMs / (1000 * 60);
+    // First turn of a session
+    if (isFirstTurn) {
+        // Check for continuation boilerplate (resumed session)
+        if (hasContinuationBoilerplate(userText)) {
+            return { label: 'continuation', confidence: 'high', source: 'continued-session' };
+        }
+        // Otherwise, definitionally a new topic
+        return { label: 'new_topic', confidence: 'high', source: 'session-boundary' };
+    }
+    // Explicit topic-shift markers (hard negatives - topic shift within session)
+    if (hasTopicShiftMarker(userText)) {
+        return { label: 'new_topic', confidence: 'high', source: 'explicit-shift-marker' };
+    }
+    // Explicit continuation markers
+    if (hasContinuationMarker(userText)) {
+        return { label: 'continuation', confidence: 'high', source: 'explicit-continuation-marker' };
+    }
+    // Large time gap suggests context switch
+    if (timeGapMinutes > timeGapThresholdMinutes) {
+        return { label: 'new_topic', confidence: 'medium', source: 'time-gap' };
+    }
+    // References to tool outputs or file paths suggest continuation
+    // (This is a simple heuristic - the lexical features module does more sophisticated detection)
+    const hasFileRef = /\.(ts|js|py|json|md|tsx|jsx|css|html|go|rs|java|c|cpp|h)\b/i.test(userText);
+    const hasToolRef = /(output|result|error|warning|log|file|path)/i.test(userText);
+    if (hasFileRef || hasToolRef) {
+        return { label: 'continuation', confidence: 'high', source: 'tool-file-reference' };
+    }
+    // Default: adjacent same-session turns are assumed to continue
+    return { label: 'continuation', confidence: 'medium', source: 'same-session-adjacent' };
+}
+/**
+ * Generate turn transitions from a single session file.
+ */
+export async function generateSessionTransitions(sessionPath, sessionId, sessionSlug, options = {}) {
+    const { timeGapMinutes = DEFAULT_TIME_GAP_MINUTES } = options;
+    const messages = await readSessionMessages(sessionPath);
+    const turns = assembleTurns(messages);
+    if (turns.length < 2) {
+        return [];
+    }
+    const transitions = [];
+    for (let i = 0; i < turns.length; i++) {
+        const nextTurn = turns[i];
+        const prevTurn = i > 0 ? turns[i - 1] : null;
+        const isFirstTurn = i === 0;
+        // Skip if user text is empty
+        if (!nextTurn.userText.trim())
+            continue;
+        // For first turn, we don't have a previous assistant output
+        // But we still want to label whether this starts a new topic
+        let prevAssistantText = '';
+        let timeGapMs = 0;
+        if (prevTurn) {
+            prevAssistantText = extractAssistantText(prevTurn.assistantBlocks);
+            const prevTime = parseTime(prevTurn.startTime);
+            const nextTime = parseTime(nextTurn.startTime);
+            timeGapMs = Math.max(0, nextTime - prevTime);
+        }
+        const { label, confidence, source } = labelTransition(prevTurn, nextTurn, isFirstTurn, timeGapMs, timeGapMinutes);
+        transitions.push({
+            id: makeTransitionId(sessionId, i - 1, i),
+            sessionId,
+            sessionSlug,
+            prevTurnIndex: isFirstTurn ? -1 : i - 1,
+            nextTurnIndex: i,
+            prevAssistantText,
+            nextUserText: nextTurn.userText,
+            timeGapMs,
+            label,
+            confidence,
+            labelSource: source,
+        });
+    }
+    return transitions;
+}
+/**
+ * Generate turn transitions from multiple sessions.
+ */
+export async function generateTransitionLabels(sessions, options = {}) {
+    const { includeLowConfidence = true } = options;
+    const allTransitions = [];
+    for (const session of sessions) {
+        try {
+            const transitions = await generateSessionTransitions(session.path, session.sessionId, session.sessionSlug, options);
+            for (const t of transitions) {
+                if (!includeLowConfidence && t.confidence === 'low') {
+                    continue;
+                }
+                allTransitions.push(t);
+            }
+            console.log(`  ${session.sessionSlug}: ${transitions.length} transitions`);
+        }
+        catch (err) {
+            console.warn(`  Failed to process ${session.path}: ${err}`);
+        }
+    }
+    return allTransitions;
+}
+/**
+ * Compute dataset statistics from transitions.
+ */
+export function computeDatasetStats(transitions) {
+    const byLabelSource = {};
+    let continuationCount = 0;
+    let newTopicCount = 0;
+    let highConfidenceCount = 0;
+    let mediumConfidenceCount = 0;
+    let lowConfidenceCount = 0;
+    for (const t of transitions) {
+        if (t.label === 'continuation')
+            continuationCount++;
+        else
+            newTopicCount++;
+        if (t.confidence === 'high')
+            highConfidenceCount++;
+        else if (t.confidence === 'medium')
+            mediumConfidenceCount++;
+        else
+            lowConfidenceCount++;
+        byLabelSource[t.labelSource] = (byLabelSource[t.labelSource] ?? 0) + 1;
+    }
+    return {
+        totalTransitions: transitions.length,
+        continuationCount,
+        newTopicCount,
+        highConfidenceCount,
+        mediumConfidenceCount,
+        lowConfidenceCount,
+        byLabelSource,
+    };
+}
+/**
+ * Filter transitions to only include high-confidence labels.
+ */
+export function filterHighConfidence(transitions) {
+    return transitions.filter((t) => t.confidence === 'high');
+}
+/**
+ * Balance dataset by undersampling the majority class.
+ */
+export function balanceDataset(transitions, seed = 42) {
+    const continuations = transitions.filter((t) => t.label === 'continuation');
+    const newTopics = transitions.filter((t) => t.label === 'new_topic');
+    const minCount = Math.min(continuations.length, newTopics.length);
+    // Deterministic shuffle
+    const shuffledContinuations = seededShuffle(continuations, seed);
+    const shuffledNewTopics = seededShuffle(newTopics, seed + 1);
+    return [
+        ...shuffledContinuations.slice(0, minCount),
+        ...shuffledNewTopics.slice(0, minCount),
+    ];
+}
+/**
+ * Deterministic seeded shuffle.
+ */
+function seededShuffle(arr, seed) {
+    const result = [...arr];
+    let s = seed;
+    for (let i = result.length - 1; i > 0; i--) {
+        s = (s * 1664525 + 1013904223) & 0x7fffffff;
+        const j = s % (i + 1);
+        [result[i], result[j]] = [result[j], result[i]];
+    }
+    return result;
+}
+//# sourceMappingURL=labeler.js.map

package/dist/eval/experiments/topic-continuity/labeler.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"labeler.js","sourceRoot":"","sources":["../../../../src/eval/experiments/topic-continuity/labeler.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,EAAE,mBAAmB,EAAE,MAAM,mCAAmC,CAAC;AACxE,OAAO,EAAE,aAAa,EAAE,MAAM,mCAAmC,CAAC;AASlE,yDAAyD;AACzD,MAAM,wBAAwB,GAAG,8DAA8D,CAAC;AAEhG,gDAAgD;AAChD,MAAM,oBAAoB,GAAG;IAC3B,0CAA0C;IAC1C,oDAAoD;IACpD,0BAA0B;IAC1B,wDAAwD;IACxD,6BAA6B;IAC7B,eAAe;IACf,mCAAmC;IACnC,iCAAiC;IACjC,kDAAkD;IAClD,oBAAoB;CACrB,CAAC;AAEF,8DAA8D;AAC9D,MAAM,qBAAqB,GAAG;IAC5B,kCAAkC;IAClC,kDAAkD;IAClD,mDAAmD;IACnD,wBAAwB;IACxB,oCAAoC;IACpC,iCAAiC;IACjC,kCAAkC;IAClC,8BAA8B;IAC9B,6CAA6C;IAC7C,mCAAmC;CACpC,CAAC;AAEF,uEAAuE;AACvE,MAAM,wBAAwB,GAAG,EAAE,CAAC;AASpC;;GAEG;AACH,SAAS,oBAAoB,CAAC,MAAsB;IAClD,OAAO,MAAM;SACV,MAAM,CAAC,CAAC,CAAC,EAAkB,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,MAAM,CAAC;SAChD,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC;SAClB,IAAI,CAAC,IAAI,CAAC,CAAC;AAChB,CAAC;AAED;;GAEG;AACH,SAAS,0BAA0B,CAAC,IAAY;IAC9C,OAAO,IAAI,CAAC,QAAQ,CAAC,wBAAwB,CAAC,CAAC;AACjD,CAAC;AAED;;GAEG;AACH,SAAS,mBAAmB,CAAC,IAAY;IACvC,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;IAC5B,OAAO,oBAAoB,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;AAC3D,CAAC;AAED;;GAEG;AACH,SAAS,qBAAqB,CAAC,IAAY;IACzC,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;IAC5B,OAAO,qBAAqB,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;AAC5D,CAAC;AAED;;GAEG;AACH,SAAS,SAAS,CAAC,GAAW;IAC5B,OAAO,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,OAAO,EAAE,CAAC;AACjC,CAAC;AAED;;GAEG;AACH,SAAS,gBAAgB,CAAC,SAAiB,EAAE,OAAe,EAAE,OAAe;IAC3E,MAAM,YAAY,GAAG,SAAS,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;IAC3C,OAAO,MAAM,YAAY,IAAI,OAAO,IAAI,OAAO,EAAE,CAAC;AACpD,CAAC;AAED;;GAEG;AACH,SAAS,eAAe,CACtB,QAAqB,EACrB,QAAc,EACd,WAAoB,EACpB,SAAiB,EACjB,uBAA+B;IAE/B,MAAM,QAAQ,GAAG,QAAQ,CAAC,QAAQ,CAAC;IACnC,MAAM,cAAc,GAAG,SAAS,GAAG,CAAC,IAAI,GAAG,EAAE,CAAC,CAAC;IAE/C,0BAA0B;IAC1B,IAAI,WAAW,EAAE,CAAC;QAChB,uDAAuD;QACvD,IAAI,0BAA0B,CAAC,QAAQ,CAAC,EAAE,CAAC;YACzC,OAAO,EAAE,KAAK,EAAE,cAAc,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,mBAAmB,EAAE,CAAC;QACpF,CAAC;QACD,wCAAwC;QACxC,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,kBAAkB,EAAE,CAAC;IAChF,CAAC;IAED,6EAA6E;IAC7E,IAAI,mBAAmB,CAAC,QAAQ,CAAC,EAAE,CAAC;QAClC,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,uBAAuB,EAAE,CAAC;IACrF,CAAC;IAED,gCAAgC;IAChC,IAAI,qBAAqB,CAAC,QAAQ,CAAC,EAAE,CAAC;QACpC,OAAO,EAAE,KAAK,EAAE,cAAc,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,8BAA8B,EAAE,CAAC;IAC/F,CAAC;IAED,yCAAyC;IACzC,IAAI,cAAc,GAAG,uBAAuB,EAAE,CAAC;QAC7C,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,UAAU,EAAE,QAAQ,EAAE,MAAM,EAAE,UAAU,EAAE,CAAC;IAC1E,CAAC;IAED,gEAAgE;IAChE,+FAA+F;IAC/F,MAAM,UAAU,GAAG,6DAA6D,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IAChG,MAAM,UAAU,GAAG,8CAA8C,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IACjF,IAAI,UAAU,IAAI,UAAU,EAAE,CAAC;QAC7B,OAAO,EAAE,KAAK,EAAE,cAAc,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,qBAAqB,EAAE,CAAC;IACtF,CAAC;IAED,+DAA+D;IAC/D,OAAO,EAAE,KAAK,EAAE,cAAc,EAAE,UAAU,EAAE,QAAQ,EAAE,MAAM,EAAE,uBAAuB,EAAE,CAAC;AAC1F,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,0BAA0B,CAC9C,WAAmB,EACnB,SAAiB,EACjB,WAAmB,EACnB,UAA0B,EAAE;IAE5B,MAAM,EAAE,cAAc,GAAG,wBAAwB,EAAE,GAAG,OAAO,CAAC;IAE9D,MAAM,QAAQ,GAAG,MAAM,mBAAmB,CAAC,WAAW,CAAC,CAAC;IACxD,MAAM,KAAK,GAAG,aAAa,CAAC,QAAQ,CAAC,CAAC;IAEtC,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACrB,OAAO,EAAE,CAAC;IACZ,CAAC;IAED,MAAM,WAAW,GAAqB,EAAE,CAAC;IAEzC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACtC,MAAM,QAAQ,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;QAC1B,MAAM,QAAQ,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;QAC7C,MAAM,WAAW,GAAG,CAAC,KAAK,CAAC,CAAC;QAE5B,6BAA6B;QAC7B,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,EAAE;YAAE,SAAS;QAExC,4DAA4D;QAC5D,6DAA6D;QAC7D,IAAI,iBAAiB,GAAG,EAAE,CAAC;QAC3B,IAAI,SAAS,GAAG,CAAC,CAAC;QAElB,IAAI,QAAQ,EAAE,CAAC;YACb,iBAAiB,GAAG,oBAAoB,CAAC,QAAQ,CAAC,eAAe,CAAC,CAAC;YACnE,MAAM,QAAQ,GAAG,SAAS,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC;YAC/C,MAAM,QAAQ,GAAG,SAAS,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC;YAC/C,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,QAAQ,GAAG,QAAQ,CAAC,CAAC;QAC/C,CAAC;QAED,MAAM,EAAE,KAAK,EAAE,UAAU,EAAE,MAAM,EAAE,GAAG,eAAe,CACnD,QAAQ,EACR,QAAQ,EACR,WAAW,EACX,SAAS,EACT,cAAc,CACf,CAAC;QAEF,WAAW,CAAC,IAAI,CAAC;YACf,EAAE,EAAE,gBAAgB,CAAC,SAAS,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;YACzC,SAAS;YACT,WAAW;YACX,aAAa,EAAE,WAAW,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC;YACvC,aAAa,EAAE,CAAC;YAChB,iBAAiB;YACjB,YAAY,EAAE,QAAQ,CAAC,QAAQ;YAC/B,SAAS;YACT,KAAK;YACL,UAAU;YACV,WAAW,EAAE,MAAM;SACpB,CAAC,CAAC;IACL,CAAC;IAED,OAAO,WAAW,CAAC;AACrB,CAAC;AAWD;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,wBAAwB,CAC5C,QAAyB,EACzB,UAA0B,EAAE;IAE5B,MAAM,EAAE,oBAAoB,GAAG,IAAI,EAAE,GAAG,OAAO,CAAC;IAEhD,MAAM,cAAc,GAAqB,EAAE,CAAC;IAE5C,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAC/B,IAAI,CAAC;YACH,MAAM,WAAW,GAAG,MAAM,0BAA0B,CAClD,OAAO,CAAC,IAAI,EACZ,OAAO,CAAC,SAAS,EACjB,OAAO,CAAC,WAAW,EACnB,OAAO,CACR,CAAC;YAEF,KAAK,MAAM,CAAC,IAAI,WAAW,EAAE,CAAC;gBAC5B,IAAI,CAAC,oBAAoB,IAAI,CAAC,CAAC,UAAU,KAAK,KAAK,EAAE,CAAC;oBACpD,SAAS;gBACX,CAAC;gBACD,cAAc,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;YACzB,CAAC;YAED,OAAO,CAAC,GAAG,CACT,KAAK,OAAO,CAAC,WAAW,KAAK,WAAW,CAAC,MAAM,cAAc,CAC9D,CAAC;QACJ,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,OAAO,CAAC,IAAI,CAAC,uBAAuB,OAAO,CAAC,IAAI,KAAK,GAAG,EAAE,CAAC,CAAC;QAC9D,CAAC;IACH,CAAC;IAED,OAAO,cAAc,CAAC;AACxB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,mBAAmB,CAAC,WAA6B;IAC/D,MAAM,aAAa,GAA2B,EAAE,CAAC;IAEjD,IAAI,iBAAiB,GAAG,CAAC,CAAC;IAC1B,IAAI,aAAa,GAAG,CAAC,CAAC;IACtB,IAAI,mBAAmB,GAAG,CAAC,CAAC;IAC5B,IAAI,qBAAqB,GAAG,CAAC,CAAC;IAC9B,IAAI,kBAAkB,GAAG,CAAC,CAAC;IAE3B,KAAK,MAAM,CAAC,IAAI,WAAW,EAAE,CAAC;QAC5B,IAAI,CAAC,CAAC,KAAK,KAAK,cAAc;YAAE,iBAAiB,EAAE,CAAC;;YAC/C,aAAa,EAAE,CAAC;QAErB,IAAI,CAAC,CAAC,UAAU,KAAK,MAAM;YAAE,mBAAmB,EAAE,CAAC;aAC9C,IAAI,CAAC,CAAC,UAAU,KAAK,QAAQ;YAAE,qBAAqB,EAAE,CAAC;;YACvD,kBAAkB,EAAE,CAAC;QAE1B,aAAa,CAAC,CAAC,CAAC,WAAW,CAAC,GAAG,CAAC,aAAa,CAAC,CAAC,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC;IACzE,CAAC;IAED,OAAO;QACL,gBAAgB,EAAE,WAAW,CAAC,MAAM;QACpC,iBAAiB;QACjB,aAAa;QACb,mBAAmB;QACnB,qBAAqB;QACrB,kBAAkB;QAClB,aAAa;KACd,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,oBAAoB,CAAC,WAA6B;IAChE,OAAO,WAAW,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,UAAU,KAAK,MAAM,CAAC,CAAC;AAC5D,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,cAAc,CAC5B,WAA6B,EAC7B,OAAe,EAAE;IAEjB,MAAM,aAAa,GAAG,WAAW,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,KAAK,cAAc,CAAC,CAAC;IAC5E,MAAM,SAAS,GAAG,WAAW,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,KAAK,WAAW,CAAC,CAAC;IAErE,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,aAAa,CAAC,MAAM,EAAE,SAAS,CAAC,MAAM,CAAC,CAAC;IAElE,wBAAwB;IACxB,MAAM,qBAAqB,GAAG,aAAa,CAAC,aAAa,EAAE,IAAI,CAAC,CAAC;IACjE,MAAM,iBAAiB,GAAG,aAAa,CAAC,SAAS,EAAE,IAAI,GAAG,CAAC,CAAC,CAAC;IAE7D,OAAO;QACL,GAAG,qBAAqB,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC;QAC3C,GAAG,iBAAiB,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC;KACxC,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,SAAS,aAAa,CAAI,GAAQ,EAAE,IAAY;IAC9C,MAAM,MAAM,GAAG,CAAC,GAAG,GAAG,CAAC,CAAC;IACxB,IAAI,CAAC,GAAG,IAAI,CAAC;IACb,KAAK,IAAI,CAAC,GAAG,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;QAC3C,CAAC,GAAG,CAAC,CAAC,GAAG,OAAO,GAAG,UAAU,CAAC,GAAG,UAAU,CAAC;QAC5C,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QACtB,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;IAClD,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC"}

package/dist/eval/experiments/topic-continuity/lexical-features.d.ts

@@ -0,0 +1,26 @@
+/**
+ * Lexical feature extraction for topic continuity classification.
+ *
+ * Re-exports shared functions from src/core/lexical-features.ts and adds
+ * eval-specific functions that depend on TransitionFeatures.
+ */
+import type { TransitionFeatures } from './types.js';
+export { hasTopicShiftMarker, hasContinuationMarker, extractFilePaths, computeFilePathOverlap, extractKeywords, computeKeywordOverlap, } from '../../../core/lexical-features.js';
+/**
+ * Extract all lexical features for a transition.
+ */
+export declare function extractLexicalFeatures(prevAssistantText: string, nextUserText: string, timeGapMs: number): Omit<TransitionFeatures, 'embeddingDistanceMin' | 'embeddingDistanceMean'>;
+/**
+ * Compute a lexical-only continuation score.
+ * Higher score = more likely to be a continuation.
+ */
+export declare function computeLexicalScore(features: Omit<TransitionFeatures, 'embeddingDistanceMin' | 'embeddingDistanceMean'>): number;
+export interface LexicalClassificationResult {
+    score: number;
+    features: Omit<TransitionFeatures, 'embeddingDistanceMin' | 'embeddingDistanceMean'>;
+}
+/**
+ * Classify a transition using only lexical features.
+ */
+export declare function classifyWithLexicalFeatures(prevAssistantText: string, nextUserText: string, timeGapMs: number): LexicalClassificationResult;
+//# sourceMappingURL=lexical-features.d.ts.map

package/dist/eval/experiments/topic-continuity/lexical-features.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"lexical-features.d.ts","sourceRoot":"","sources":["../../../../src/eval/experiments/topic-continuity/lexical-features.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,KAAK,EAAE,kBAAkB,EAAE,MAAM,YAAY,CAAC;AASrD,OAAO,EACL,mBAAmB,EACnB,qBAAqB,EACrB,gBAAgB,EAChB,sBAAsB,EACtB,eAAe,EACf,qBAAqB,GACtB,MAAM,mCAAmC,CAAC;AAE3C;;GAEG;AACH,wBAAgB,sBAAsB,CACpC,iBAAiB,EAAE,MAAM,EACzB,YAAY,EAAE,MAAM,EACpB,SAAS,EAAE,MAAM,GAChB,IAAI,CAAC,kBAAkB,EAAE,sBAAsB,GAAG,uBAAuB,CAAC,CAQ5E;AAED;;;GAGG;AACH,wBAAgB,mBAAmB,CACjC,QAAQ,EAAE,IAAI,CAAC,kBAAkB,EAAE,sBAAsB,GAAG,uBAAuB,CAAC,GACnF,MAAM,CA4BR;AAED,MAAM,WAAW,2BAA2B;IAC1C,KAAK,EAAE,MAAM,CAAC;IACd,QAAQ,EAAE,IAAI,CAAC,kBAAkB,EAAE,sBAAsB,GAAG,uBAAuB,CAAC,CAAC;CACtF;AAED;;GAEG;AACH,wBAAgB,2BAA2B,CACzC,iBAAiB,EAAE,MAAM,EACzB,YAAY,EAAE,MAAM,EACpB,SAAS,EAAE,MAAM,GAChB,2BAA2B,CAI7B"}

package/dist/eval/experiments/topic-continuity/lexical-features.js

@@ -0,0 +1,58 @@
+/**
+ * Lexical feature extraction for topic continuity classification.
+ *
+ * Re-exports shared functions from src/core/lexical-features.ts and adds
+ * eval-specific functions that depend on TransitionFeatures.
+ */
+import { hasTopicShiftMarker as coreHasTopicShiftMarker, hasContinuationMarker as coreHasContinuationMarker, computeFilePathOverlap as coreComputeFilePathOverlap, computeKeywordOverlap as coreComputeKeywordOverlap, } from '../../../core/lexical-features.js';
+// Re-export all shared functions from core
+export { hasTopicShiftMarker, hasContinuationMarker, extractFilePaths, computeFilePathOverlap, extractKeywords, computeKeywordOverlap, } from '../../../core/lexical-features.js';
+/**
+ * Extract all lexical features for a transition.
+ */
+export function extractLexicalFeatures(prevAssistantText, nextUserText, timeGapMs) {
+    return {
+        timeGapMinutes: timeGapMs / (1000 * 60),
+        hasTopicShiftMarker: coreHasTopicShiftMarker(nextUserText),
+        hasContinuationMarker: coreHasContinuationMarker(nextUserText),
+        filePathOverlap: coreComputeFilePathOverlap(prevAssistantText, nextUserText),
+        keywordOverlap: coreComputeKeywordOverlap(prevAssistantText, nextUserText),
+    };
+}
+/**
+ * Compute a lexical-only continuation score.
+ * Higher score = more likely to be a continuation.
+ */
+export function computeLexicalScore(features) {
+    let score = 0.5; // Start at neutral
+    // Topic shift markers strongly indicate new topic
+    if (features.hasTopicShiftMarker) {
+        score -= 0.4;
+    }
+    // Continuation markers strongly indicate continuation
+    if (features.hasContinuationMarker) {
+        score += 0.3;
+    }
+    // File path overlap suggests continuation
+    score += features.filePathOverlap * 0.2;
+    // Keyword overlap suggests continuation
+    score += features.keywordOverlap * 0.15;
+    // Large time gap suggests new topic
+    if (features.timeGapMinutes > 30) {
+        score -= 0.25;
+    }
+    else if (features.timeGapMinutes > 10) {
+        score -= 0.1;
+    }
+    // Clamp to [0, 1]
+    return Math.max(0, Math.min(1, score));
+}
+/**
+ * Classify a transition using only lexical features.
+ */
+export function classifyWithLexicalFeatures(prevAssistantText, nextUserText, timeGapMs) {
+    const features = extractLexicalFeatures(prevAssistantText, nextUserText, timeGapMs);
+    const score = computeLexicalScore(features);
+    return { score, features };
+}
+//# sourceMappingURL=lexical-features.js.map

package/dist/eval/experiments/topic-continuity/lexical-features.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"lexical-features.js","sourceRoot":"","sources":["../../../../src/eval/experiments/topic-continuity/lexical-features.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAGH,OAAO,EACL,mBAAmB,IAAI,uBAAuB,EAC9C,qBAAqB,IAAI,yBAAyB,EAClD,sBAAsB,IAAI,0BAA0B,EACpD,qBAAqB,IAAI,yBAAyB,GACnD,MAAM,mCAAmC,CAAC;AAE3C,2CAA2C;AAC3C,OAAO,EACL,mBAAmB,EACnB,qBAAqB,EACrB,gBAAgB,EAChB,sBAAsB,EACtB,eAAe,EACf,qBAAqB,GACtB,MAAM,mCAAmC,CAAC;AAE3C;;GAEG;AACH,MAAM,UAAU,sBAAsB,CACpC,iBAAyB,EACzB,YAAoB,EACpB,SAAiB;IAEjB,OAAO;QACL,cAAc,EAAE,SAAS,GAAG,CAAC,IAAI,GAAG,EAAE,CAAC;QACvC,mBAAmB,EAAE,uBAAuB,CAAC,YAAY,CAAC;QAC1D,qBAAqB,EAAE,yBAAyB,CAAC,YAAY,CAAC;QAC9D,eAAe,EAAE,0BAA0B,CAAC,iBAAiB,EAAE,YAAY,CAAC;QAC5E,cAAc,EAAE,yBAAyB,CAAC,iBAAiB,EAAE,YAAY,CAAC;KAC3E,CAAC;AACJ,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,mBAAmB,CACjC,QAAoF;IAEpF,IAAI,KAAK,GAAG,GAAG,CAAC,CAAC,mBAAmB;IAEpC,kDAAkD;IAClD,IAAI,QAAQ,CAAC,mBAAmB,EAAE,CAAC;QACjC,KAAK,IAAI,GAAG,CAAC;IACf,CAAC;IAED,sDAAsD;IACtD,IAAI,QAAQ,CAAC,qBAAqB,EAAE,CAAC;QACnC,KAAK,IAAI,GAAG,CAAC;IACf,CAAC;IAED,0CAA0C;IAC1C,KAAK,IAAI,QAAQ,CAAC,eAAe,GAAG,GAAG,CAAC;IAExC,wCAAwC;IACxC,KAAK,IAAI,QAAQ,CAAC,cAAc,GAAG,IAAI,CAAC;IAExC,oCAAoC;IACpC,IAAI,QAAQ,CAAC,cAAc,GAAG,EAAE,EAAE,CAAC;QACjC,KAAK,IAAI,IAAI,CAAC;IAChB,CAAC;SAAM,IAAI,QAAQ,CAAC,cAAc,GAAG,EAAE,EAAE,CAAC;QACxC,KAAK,IAAI,GAAG,CAAC;IACf,CAAC;IAED,kBAAkB;IAClB,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC;AACzC,CAAC;AAOD;;GAEG;AACH,MAAM,UAAU,2BAA2B,CACzC,iBAAyB,EACzB,YAAoB,EACpB,SAAiB;IAEjB,MAAM,QAAQ,GAAG,sBAAsB,CAAC,iBAAiB,EAAE,YAAY,EAAE,SAAS,CAAC,CAAC;IACpF,MAAM,KAAK,GAAG,mBAAmB,CAAC,QAAQ,CAAC,CAAC;IAC5C,OAAO,EAAE,KAAK,EAAE,QAAQ,EAAE,CAAC;AAC7B,CAAC"}