@hongmaple0820/med-scale-research-os 0.43.0

package/dist/knowledge/KnowledgeBase.js

@@ -0,0 +1,182 @@
+// SCALE Engine — Knowledge Base (W7 完整实现 + v0.7.2 TF-IDF)
+// 设计参考：docs/03-CORE-MODULES.md §3.4
+// Phase 3 增强：TF-IDF 文本相似度计算
+export class KnowledgeBase {
+    constructor(eventBus) {
+        this.eventBus = eventBus;
+        this.entries = new Map();
+        this.seq = 0;
+        // TF-IDF cache
+        this.documentFrequencies = new Map();
+        this.totalDocuments = 0;
+    }
+    async add(input) {
+        const entry = {
+            ...input,
+            id: this.generateId(),
+            createdAt: Date.now(),
+            accessCount: 0,
+            relevance: 0.5,
+        };
+        this.entries.set(entry.id, entry);
+        this.eventBus.emit('lesson.proposed', { lessonId: entry.id }, { artifactId: input.sourceArtifact });
+        return entry;
+    }
+    async recall(query) {
+        let results = Array.from(this.entries.values());
+        if (query.type) {
+            const types = Array.isArray(query.type) ? query.type : [query.type];
+            results = results.filter((e) => types.includes(e.type));
+        }
+        if (query.tags)
+            results = results.filter((e) => query.tags.every((t) => e.tags.includes(t)));
+        if (query.minRelevance)
+            results = results.filter((e) => e.relevance >= query.minRelevance);
+        if (query.verifiedOnly)
+            results = results.filter((e) => e.verified);
+        results.sort((a, b) => b.relevance - a.relevance);
+        return results.slice(0, query.limit ?? 10);
+    }
+    async recallByVector(text, topK) {
+        // v0.7.2: TF-IDF implementation (Phase 3 enhancement)
+        const queryTerms = this.tokenize(text);
+        if (queryTerms.length === 0)
+            return this.recall({ verifiedOnly: true, limit: topK });
+        const scored = [];
+        for (const entry of this.entries.values()) {
+            // Use title + tags for document representation (contentRef is file path)
+            const docText = `${entry.title} ${entry.tags.join(' ')}`;
+            const docTerms = this.tokenize(docText);
+            const score = this.cosineSimilarity(queryTerms, docTerms);
+            if (score > 0)
+                scored.push({ entry, score });
+        }
+        // Fallback to verified recall if no TF-IDF matches
+        if (scored.length === 0)
+            return this.recall({ verifiedOnly: true, limit: topK });
+        // Sort by TF-IDF similarity, then by relevance as tiebreaker
+        scored.sort((a, b) => b.score - a.score || b.entry.relevance - a.entry.relevance);
+        return scored.slice(0, topK).map(s => s.entry);
+    }
+    // ============================================================================
+    // TF-IDF Helper Methods (Phase 3)
+    // ============================================================================
+    /**
+     * Tokenize text into lowercase terms (simple word splitting)
+     */
+    tokenize(text) {
+        return text
+            .toLowerCase()
+            .replace(/[^\w\s]/g, ' ')
+            .split(/\s+/)
+            .filter(t => t.length > 2); // Skip short terms
+    }
+    /**
+     * Calculate term frequency (TF) for a term in a document
+     */
+    termFrequency(term, docTerms) {
+        const count = docTerms.filter(t => t === term).length;
+        return count / docTerms.length;
+    }
+    /**
+     * Calculate inverse document frequency (IDF) for a term
+     */
+    inverseDocumentFrequency(term) {
+        // Update document frequencies if needed
+        if (this.documentFrequencies.size === 0 || this.totalDocuments !== this.entries.size) {
+            this.rebuildDocumentFrequencies();
+        }
+        const df = this.documentFrequencies.get(term) ?? 0;
+        if (df === 0)
+            return 0;
+        return Math.log(this.totalDocuments / df);
+    }
+    /**
+     * Rebuild document frequency cache
+     */
+    rebuildDocumentFrequencies() {
+        this.documentFrequencies.clear();
+        this.totalDocuments = this.entries.size;
+        for (const entry of this.entries.values()) {
+            // Use title + tags for document representation (contentRef is file path)
+            const docText = `${entry.title} ${entry.tags.join(' ')}`;
+            const terms = new Set(this.tokenize(docText));
+            for (const term of terms) {
+                this.documentFrequencies.set(term, (this.documentFrequencies.get(term) ?? 0) + 1);
+            }
+        }
+    }
+    /**
+     * Calculate TF-IDF vector for a document
+     */
+    tfidfVector(terms) {
+        const tfidf = new Map();
+        const uniqueTerms = new Set(terms);
+        for (const term of uniqueTerms) {
+            const tf = this.termFrequency(term, terms);
+            const idf = this.inverseDocumentFrequency(term);
+            if (tf * idf > 0)
+                tfidf.set(term, tf * idf);
+        }
+        return tfidf;
+    }
+    /**
+     * Calculate cosine similarity between two term sets
+     */
+    cosineSimilarity(queryTerms, docTerms) {
+        const queryVec = this.tfidfVector(queryTerms);
+        const docVec = this.tfidfVector(docTerms);
+        if (queryVec.size === 0 || docVec.size === 0)
+            return 0;
+        // Dot product
+        let dot = 0;
+        for (const [term, weight] of queryVec) {
+            if (docVec.has(term))
+                dot += weight * docVec.get(term);
+        }
+        // Magnitudes
+        const queryMag = Math.sqrt(Array.from(queryVec.values()).reduce((s, w) => s + w * w, 0));
+        const docMag = Math.sqrt(Array.from(docVec.values()).reduce((s, w) => s + w * w, 0));
+        if (queryMag === 0 || docMag === 0)
+            return 0;
+        return dot / (queryMag * docMag);
+    }
+    async markHelpful(id, sessionId) {
+        const entry = this.entries.get(id);
+        if (entry) {
+            entry.relevance = Math.min(1, entry.relevance + 0.05);
+            entry.accessCount += 1;
+            entry.lastAccessed = Date.now();
+            this.eventBus.emit('lesson.helpful', { lessonId: id }, { sessionId });
+        }
+    }
+    async markUseless(id, sessionId) {
+        const entry = this.entries.get(id);
+        if (entry) {
+            entry.relevance = Math.max(0.05, entry.relevance - 0.1);
+            this.eventBus.emit('lesson.useless', { lessonId: id }, { sessionId });
+        }
+    }
+    async verify(id, verifiedBy) {
+        const entry = this.entries.get(id);
+        if (entry) {
+            entry.verified = true;
+            entry.verifiedBy = verifiedBy;
+            entry.verifiedAt = Date.now();
+            this.eventBus.emit('lesson.approved', { lessonId: id, verifiedBy });
+        }
+    }
+    async decay() {
+        const DAY = 24 * 60 * 60 * 1000;
+        for (const entry of this.entries.values()) {
+            const days = entry.lastAccessed ? (Date.now() - entry.lastAccessed) / DAY : 90;
+            const recency = Math.exp(-days / 30);
+            entry.relevance = Math.max(0.05, entry.relevance * 0.95 + recency * 0.05);
+        }
+    }
+    generateId() {
+        this.seq = (this.seq + 1) % 10000;
+        return `KB-${Date.now()}-${this.seq.toString().padStart(4, '0')}`;
+    }
+}
+//# sourceMappingURL=KnowledgeBase.js.map

package/dist/knowledge/KnowledgeBase.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"KnowledgeBase.js","sourceRoot":"","sources":["../../src/knowledge/KnowledgeBase.ts"],"names":[],"mappings":"AAAA,0DAA0D;AAC1D,oCAAoC;AACpC,4BAA4B;AAgB5B,MAAM,OAAO,aAAa;IAOxB,YAAoB,QAAmB;QAAnB,aAAQ,GAAR,QAAQ,CAAW;QAN/B,YAAO,GAAG,IAAI,GAAG,EAA0B,CAAA;QAC3C,QAAG,GAAG,CAAC,CAAA;QACf,eAAe;QACP,wBAAmB,GAAG,IAAI,GAAG,EAAkB,CAAA;QAC/C,mBAAc,GAAG,CAAC,CAAA;IAEgB,CAAC;IAE3C,KAAK,CAAC,GAAG,CAAC,KAA6E;QACrF,MAAM,KAAK,GAAmB;YAC5B,GAAG,KAAK;YACR,EAAE,EAAE,IAAI,CAAC,UAAU,EAAE;YACrB,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE;YACrB,WAAW,EAAE,CAAC;YACd,SAAS,EAAE,GAAG;SACf,CAAA;QACD,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,EAAE,KAAK,CAAC,CAAA;QACjC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,iBAAiB,EAAE,EAAE,QAAQ,EAAE,KAAK,CAAC,EAAE,EAAE,EAAE,EAAE,UAAU,EAAE,KAAK,CAAC,cAAc,EAAE,CAAC,CAAA;QACnG,OAAO,KAAK,CAAA;IACd,CAAC;IAED,KAAK,CAAC,MAAM,CAAC,KAAqB;QAChC,IAAI,OAAO,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC,CAAA;QAC/C,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC;YACf,MAAM,KAAK,GAAG,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,CAAA;YACnE,OAAO,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAA;QACzD,CAAC;QACD,IAAI,KAAK,CAAC,IAAI;YAAE,OAAO,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,KAAK,CAAC,IAAK,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;QAC7F,IAAI,KAAK,CAAC,YAAY;YAAE,OAAO,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,SAAS,IAAI,KAAK,CAAC,YAAa,CAAC,CAAA;QAC3F,IAAI,KAAK,CAAC,YAAY;YAAE,OAAO,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAA;QACnE,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,SAAS,GAAG,CAAC,CAAC,SAAS,CAAC,CAAA;QACjD,OAAO,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,KAAK,IAAI,EAAE,CAAC,CAAA;IAC5C,CAAC;IAED,KAAK,CAAC,cAAc,CAAC,IAAY,EAAE,IAAY;QAC7C,sDAAsD;QACtD,MAAM,UAAU,GAAG,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAA;QACtC,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,IAAI,CAAC,MAAM,CAAC,EAAE,YAAY,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAA;QAEpF,MAAM,MAAM,GAAoD,EAAE,CAAA;QAElE,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC;YAC1C,yEAAyE;YACzE,MAAM,OAAO,GAAG,GAAG,KAAK,CAAC,KAAK,IAAI,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAA;YACxD,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAA;YACvC,MAAM,KAAK,GAAG,IAAI,CAAC,gBAAgB,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAA;YACzD,IAAI,KAAK,GAAG,CAAC;gBAAE,MAAM,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC,CAAA;QAC9C,CAAC;QAED,mDAAmD;QACnD,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,IAAI,CAAC,MAAM,CAAC,EAAE,YAAY,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAA;QAEhF,6DAA6D;QAC7D,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,KAAK,CAAC,SAAS,GAAG,CAAC,CAAC,KAAK,CAAC,SAAS,CAAC,CAAA;QAEjF,OAAO,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAA;IAChD,CAAC;IAED,+EAA+E;IAC/E,kCAAkC;IAClC,+EAA+E;IAE/E;;OAEG;IACK,QAAQ,CAAC,IAAY;QAC3B,OAAO,IAAI;aACR,WAAW,EAAE;aACb,OAAO,CAAC,UAAU,EAAE,GAAG,CAAC;aACxB,KAAK,CAAC,KAAK,CAAC;aACZ,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA,CAAC,mBAAmB;IAClD,CAAC;IAED;;OAEG;IACK,aAAa,CAAC,IAAY,EAAE,QAAkB;QACpD,MAAM,KAAK,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,MAAM,CAAA;QACrD,OAAO,KAAK,GAAG,QAAQ,CAAC,MAAM,CAAA;IAChC,CAAC;IAED;;OAEG;IACK,wBAAwB,CAAC,IAAY;QAC3C,wCAAwC;QACxC,IAAI,IAAI,CAAC,mBAAmB,CAAC,IAAI,KAAK,CAAC,IAAI,IAAI,CAAC,cAAc,KAAK,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC;YACrF,IAAI,CAAC,0BAA0B,EAAE,CAAA;QACnC,CAAC;QACD,MAAM,EAAE,GAAG,IAAI,CAAC,mBAAmB,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QAClD,IAAI,EAAE,KAAK,CAAC;YAAE,OAAO,CAAC,CAAA;QACtB,OAAO,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,cAAc,GAAG,EAAE,CAAC,CAAA;IAC3C,CAAC;IAED;;OAEG;IACK,0BAA0B;QAChC,IAAI,CAAC,mBAAmB,CAAC,KAAK,EAAE,CAAA;QAChC,IAAI,CAAC,cAAc,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,CAAA;QAEvC,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC;YAC1C,yEAAyE;YACzE,MAAM,OAAO,GAAG,GAAG,KAAK,CAAC,KAAK,IAAI,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAA;YACxD,MAAM,KAAK,GAAG,IAAI,GAAG,CAAC,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAC,CAAA;YAC7C,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;gBACzB,IAAI,CAAC,mBAAmB,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,IAAI,CAAC,mBAAmB,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAA;YACnF,CAAC;QACH,CAAC;IACH,CAAC;IAED;;OAEG;IACK,WAAW,CAAC,KAAe;QACjC,MAAM,KAAK,GAAG,IAAI,GAAG,EAAkB,CAAA;QACvC,MAAM,WAAW,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAA;QAClC,KAAK,MAAM,IAAI,IAAI,WAAW,EAAE,CAAC;YAC/B,MAAM,EAAE,GAAG,IAAI,CAAC,aAAa,CAAC,IAAI,EAAE,KAAK,CAAC,CAAA;YAC1C,MAAM,GAAG,GAAG,IAAI,CAAC,wBAAwB,CAAC,IAAI,CAAC,CAAA;YAC/C,IAAI,EAAE,GAAG,GAAG,GAAG,CAAC;gBAAE,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,EAAE,GAAG,GAAG,CAAC,CAAA;QAC7C,CAAC;QACD,OAAO,KAAK,CAAA;IACd,CAAC;IAED;;OAEG;IACK,gBAAgB,CAAC,UAAoB,EAAE,QAAkB;QAC/D,MAAM,QAAQ,GAAG,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,CAAA;QAC7C,MAAM,MAAM,GAAG,IAAI,CAAC,WAAW,CAAC,QAAQ,CAAC,CAAA;QAEzC,IAAI,QAAQ,CAAC,IAAI,KAAK,CAAC,IAAI,MAAM,CAAC,IAAI,KAAK,CAAC;YAAE,OAAO,CAAC,CAAA;QAEtD,cAAc;QACd,IAAI,GAAG,GAAG,CAAC,CAAA;QACX,KAAK,MAAM,CAAC,IAAI,EAAE,MAAM,CAAC,IAAI,QAAQ,EAAE,CAAC;YACtC,IAAI,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC;gBAAE,GAAG,IAAI,MAAM,GAAG,MAAM,CAAC,GAAG,CAAC,IAAI,CAAE,CAAA;QACzD,CAAC;QAED,aAAa;QACb,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,MAAM,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QACxF,MAAM,MAAM,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QAEpF,IAAI,QAAQ,KAAK,CAAC,IAAI,MAAM,KAAK,CAAC;YAAE,OAAO,CAAC,CAAA;QAC5C,OAAO,GAAG,GAAG,CAAC,QAAQ,GAAG,MAAM,CAAC,CAAA;IAClC,CAAC;IAED,KAAK,CAAC,WAAW,CAAC,EAAU,EAAE,SAAiB;QAC7C,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;QAClC,IAAI,KAAK,EAAE,CAAC;YACV,KAAK,CAAC,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,SAAS,GAAG,IAAI,CAAC,CAAA;YACrD,KAAK,CAAC,WAAW,IAAI,CAAC,CAAA;YACtB,KAAK,CAAC,YAAY,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;YAC/B,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,gBAAgB,EAAE,EAAE,QAAQ,EAAE,EAAE,EAAE,EAAE,EAAE,SAAS,EAAE,CAAC,CAAA;QACvE,CAAC;IACH,CAAC;IAED,KAAK,CAAC,WAAW,CAAC,EAAU,EAAE,SAAiB;QAC7C,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;QAClC,IAAI,KAAK,EAAE,CAAC;YACV,KAAK,CAAC,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,CAAC,SAAS,GAAG,GAAG,CAAC,CAAA;YACvD,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,gBAAgB,EAAE,EAAE,QAAQ,EAAE,EAAE,EAAE,EAAE,EAAE,SAAS,EAAE,CAAC,CAAA;QACvE,CAAC;IACH,CAAC;IAED,KAAK,CAAC,MAAM,CAAC,EAAU,EAAE,UAAkB;QACzC,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;QAClC,IAAI,KAAK,EAAE,CAAC;YACV,KAAK,CAAC,QAAQ,GAAG,IAAI,CAAA;YACrB,KAAK,CAAC,UAAU,GAAG,UAAU,CAAA;YAC7B,KAAK,CAAC,UAAU,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;YAC7B,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,iBAAiB,EAAE,EAAE,QAAQ,EAAE,EAAE,EAAE,UAAU,EAAE,CAAC,CAAA;QACrE,CAAC;IACH,CAAC;IAED,KAAK,CAAC,KAAK;QACT,MAAM,GAAG,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,IAAI,CAAA;QAC/B,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC;YAC1C,MAAM,IAAI,GAAG,KAAK,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,GAAG,KAAK,CAAC,YAAY,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,EAAE,CAAA;YAC9E,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,GAAG,EAAE,CAAC,CAAA;YACpC,KAAK,CAAC,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,CAAC,SAAS,GAAG,IAAI,GAAG,OAAO,GAAG,IAAI,CAAC,CAAA;QAC3E,CAAC;IACH,CAAC;IAEO,UAAU;QAChB,IAAI,CAAC,GAAG,GAAG,CAAC,IAAI,CAAC,GAAG,GAAG,CAAC,CAAC,GAAG,KAAK,CAAA;QACjC,OAAO,MAAM,IAAI,CAAC,GAAG,EAAE,IAAI,IAAI,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,QAAQ,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,CAAA;IACnE,CAAC;CACF"}

package/dist/knowledge/SQLiteKnowledgeBase.d.ts

@@ -0,0 +1,29 @@
+import type { KnowledgeEntry, KnowledgeQuery } from '../artifact/types.js';
+import type { IEventBus } from '../core/eventBus.js';
+import type { IKnowledgeBase } from './KnowledgeBase.js';
+export declare class SQLiteKnowledgeBase implements IKnowledgeBase {
+    private eventBus;
+    private db;
+    private seq;
+    private tfidfIndex;
+    constructor(eventBus: IEventBus, opts?: {
+        dbPath?: string;
+    });
+    add(input: Omit<KnowledgeEntry, 'id' | 'createdAt' | 'accessCount' | 'relevance'>): Promise<KnowledgeEntry>;
+    recall(query: KnowledgeQuery): Promise<KnowledgeEntry[]>;
+    recallByVector(text: string, topK: number): Promise<KnowledgeEntry[]>;
+    markHelpful(id: string, sessionId: string): Promise<void>;
+    markUseless(id: string, sessionId: string): Promise<void>;
+    verify(id: string, verifiedBy: string): Promise<void>;
+    decay(): Promise<void>;
+    /** Close database connection */
+    close(): void;
+    /** Get stats */
+    stats(): {
+        entryCount: number;
+        verifiedCount: number;
+        byType: Record<string, number>;
+    };
+    private generateId;
+    private fromRow;
+}

package/dist/knowledge/SQLiteKnowledgeBase.js

@@ -0,0 +1,203 @@
+// SCALE Engine — SQLite Knowledge Base
+// Persistent version of KnowledgeBase using better-sqlite3
+// Design ref: docs/03-CORE-MODULES.md §3.4
+import Database from 'better-sqlite3';
+import { TfidfIndex } from './TfidfIndex.js';
+import { logger } from '../core/logger.js';
+import { mkdirSync } from 'node:fs';
+import { dirname } from 'node:path';
+const KB_SCHEMA = `
+CREATE TABLE IF NOT EXISTS knowledge_entries (
+  id              TEXT PRIMARY KEY,
+  type            TEXT NOT NULL,
+  title           TEXT NOT NULL,
+  tags            TEXT NOT NULL DEFAULT '[]',
+  content_ref     TEXT NOT NULL DEFAULT '',
+  embedding_id    TEXT,
+  relevance       REAL NOT NULL DEFAULT 0.5,
+  access_count    INTEGER NOT NULL DEFAULT 0,
+  last_accessed   INTEGER,
+  verified        INTEGER NOT NULL DEFAULT 0,
+  verified_by     TEXT,
+  verified_at     INTEGER,
+  created_at      INTEGER NOT NULL,
+  source_artifact TEXT
+);
+
+CREATE INDEX IF NOT EXISTS idx_kb_type ON knowledge_entries(type);
+CREATE INDEX IF NOT EXISTS idx_kb_relevance ON knowledge_entries(relevance);
+CREATE INDEX IF NOT EXISTS idx_kb_verified ON knowledge_entries(verified);
+CREATE INDEX IF NOT EXISTS idx_kb_created ON knowledge_entries(created_at);
+`;
+export class SQLiteKnowledgeBase {
+    constructor(eventBus, opts = {}) {
+        this.eventBus = eventBus;
+        this.seq = 0;
+        this.tfidfIndex = null;
+        const dbPath = opts.dbPath ?? '.scale/knowledge.db';
+        mkdirSync(dirname(dbPath), { recursive: true });
+        this.db = new Database(dbPath);
+        this.db.pragma('journal_mode = WAL');
+        this.db.pragma('busy_timeout = 5000');
+        this.db.exec(KB_SCHEMA);
+        // Restore seq from max id
+        const maxRow = this.db.prepare('SELECT id FROM knowledge_entries ORDER BY created_at DESC LIMIT 1').get();
+        if (maxRow) {
+            const parts = maxRow.id.split('-');
+            this.seq = parseInt(parts[parts.length - 1], 10) || 0;
+        }
+    }
+    async add(input) {
+        const entry = {
+            ...input,
+            id: this.generateId(),
+            createdAt: Date.now(),
+            accessCount: 0,
+            relevance: 0.5,
+        };
+        this.db.prepare(`
+      INSERT INTO knowledge_entries
+        (id, type, title, tags, content_ref, embedding_id, relevance,
+         access_count, last_accessed, verified, verified_by, verified_at,
+         created_at, source_artifact)
+      VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+    `).run(entry.id, entry.type, entry.title, JSON.stringify(entry.tags), entry.contentRef, entry.embeddingId ?? null, entry.relevance, entry.accessCount, entry.lastAccessed ?? null, entry.verified ? 1 : 0, entry.verifiedBy ?? null, entry.verifiedAt ?? null, entry.createdAt, entry.sourceArtifact ?? null);
+        this.eventBus.emit('lesson.proposed', { lessonId: entry.id }, { artifactId: input.sourceArtifact });
+        // Update TF-IDF index
+        if (this.tfidfIndex) {
+            const text = [entry.title, JSON.stringify(entry.tags), entry.contentRef].filter(Boolean).join(' ');
+            this.tfidfIndex.upsert(entry.id, text);
+        }
+        return entry;
+    }
+    async recall(query) {
+        const conditions = ['1=1'];
+        const params = [];
+        if (query.type) {
+            const types = Array.isArray(query.type) ? query.type : [query.type];
+            conditions.push(`type IN (${types.map(() => '?').join(',')})`);
+            params.push(...types);
+        }
+        if (query.tags && query.tags.length > 0) {
+            for (const tag of query.tags) {
+                conditions.push(`tags LIKE ?`);
+                params.push(`%"${tag}"%`);
+            }
+        }
+        if (query.minRelevance !== undefined) {
+            conditions.push('relevance >= ?');
+            params.push(query.minRelevance);
+        }
+        if (query.verifiedOnly) {
+            conditions.push('verified = 1');
+        }
+        const limit = query.limit ?? 10;
+        const sql = `SELECT * FROM knowledge_entries WHERE ${conditions.join(' AND ')} ORDER BY relevance DESC LIMIT ?`;
+        params.push(limit);
+        const rows = this.db.prepare(sql).all(...params);
+        return rows.map((r) => this.fromRow(r));
+    }
+    async recallByVector(text, topK) {
+        // Build TF-IDF index if not already built
+        if (!this.tfidfIndex) {
+            this.tfidfIndex = new TfidfIndex();
+            const rows = this.db.prepare('SELECT id, title, tags, content_ref FROM knowledge_entries').all();
+            this.tfidfIndex.buildFromRows(rows);
+            logger.debug({ docCount: rows.length }, 'TF-IDF index built');
+        }
+        const results = this.tfidfIndex.search(text, topK);
+        if (results.length === 0) {
+            logger.debug({ text, topK }, 'recallByVector: no matches found');
+            return this.recall({ verifiedOnly: true, limit: topK });
+        }
+        const entries = [];
+        for (const { id, score } of results) {
+            const row = this.db.prepare('SELECT * FROM knowledge_entries WHERE id = ?').get(id);
+            if (row) {
+                const entry = this.fromRow(row);
+                // Boost relevance by similarity score
+                entry.relevance = Math.min(1, entry.relevance * (1 + score));
+                entries.push(entry);
+            }
+        }
+        logger.debug({ text, topK, matches: entries.length }, 'recallByVector (TF-IDF)');
+        return entries;
+    }
+    async markHelpful(id, sessionId) {
+        const row = this.db.prepare('SELECT * FROM knowledge_entries WHERE id = ?').get(id);
+        if (!row)
+            return;
+        const newRelevance = Math.min(1, row.relevance + 0.05);
+        const newCount = row.access_count + 1;
+        const now = Date.now();
+        this.db.prepare('UPDATE knowledge_entries SET relevance = ?, access_count = ?, last_accessed = ? WHERE id = ?').run(newRelevance, newCount, now, id);
+        this.eventBus.emit('lesson.helpful', { lessonId: id }, { sessionId });
+    }
+    async markUseless(id, sessionId) {
+        const row = this.db.prepare('SELECT * FROM knowledge_entries WHERE id = ?').get(id);
+        if (!row)
+            return;
+        const newRelevance = Math.max(0.05, row.relevance - 0.1);
+        this.db.prepare('UPDATE knowledge_entries SET relevance = ? WHERE id = ?').run(newRelevance, id);
+        this.eventBus.emit('lesson.useless', { lessonId: id }, { sessionId });
+    }
+    async verify(id, verifiedBy) {
+        const now = Date.now();
+        const result = this.db.prepare('UPDATE knowledge_entries SET verified = 1, verified_by = ?, verified_at = ? WHERE id = ?').run(verifiedBy, now, id);
+        if (result.changes > 0) {
+            this.eventBus.emit('lesson.approved', { lessonId: id, verifiedBy });
+        }
+    }
+    async decay() {
+        const DAY = 24 * 60 * 60 * 1000;
+        const now = Date.now();
+        const rows = this.db.prepare('SELECT id, relevance, last_accessed FROM knowledge_entries').all();
+        const stmt = this.db.prepare('UPDATE knowledge_entries SET relevance = ? WHERE id = ?');
+        const updateAll = this.db.transaction(() => {
+            for (const row of rows) {
+                const days = row.last_accessed ? (now - row.last_accessed) / DAY : 90;
+                const recency = Math.exp(-days / 30);
+                const newRelevance = Math.max(0.05, row.relevance * 0.95 + recency * 0.05);
+                stmt.run(newRelevance, row.id);
+            }
+        });
+        updateAll();
+    }
+    /** Close database connection */
+    close() {
+        this.db.close();
+    }
+    /** Get stats */
+    stats() {
+        const entryCount = this.db.prepare('SELECT COUNT(*) as c FROM knowledge_entries').get().c;
+        const verifiedCount = this.db.prepare('SELECT COUNT(*) as c FROM knowledge_entries WHERE verified = 1').get().c;
+        const rows = this.db.prepare('SELECT type, COUNT(*) as c FROM knowledge_entries GROUP BY type').all();
+        const byType = {};
+        for (const r of rows)
+            byType[r.type] = r.c;
+        return { entryCount, verifiedCount, byType };
+    }
+    generateId() {
+        this.seq = (this.seq + 1) % 10000;
+        return `KB-${Date.now()}-${this.seq.toString().padStart(4, '0')}`;
+    }
+    fromRow(row) {
+        return {
+            id: row.id,
+            type: row.type,
+            title: row.title,
+            tags: JSON.parse(row.tags),
+            contentRef: row.content_ref,
+            embeddingId: row.embedding_id ?? undefined,
+            relevance: row.relevance,
+            accessCount: row.access_count,
+            lastAccessed: row.last_accessed ?? undefined,
+            verified: row.verified === 1,
+            verifiedBy: row.verified_by ?? undefined,
+            verifiedAt: row.verified_at ?? undefined,
+            createdAt: row.created_at,
+            sourceArtifact: row.source_artifact ?? undefined,
+        };
+    }
+}
+//# sourceMappingURL=SQLiteKnowledgeBase.js.map

package/dist/knowledge/SQLiteKnowledgeBase.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"SQLiteKnowledgeBase.js","sourceRoot":"","sources":["../../src/knowledge/SQLiteKnowledgeBase.ts"],"names":[],"mappings":"AAAA,uCAAuC;AACvC,2DAA2D;AAC3D,2CAA2C;AAE3C,OAAO,QAAQ,MAAM,gBAAgB,CAAA;AAIrC,OAAO,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAA;AAC5C,OAAO,EAAE,MAAM,EAAE,MAAM,mBAAmB,CAAA;AAC1C,OAAO,EAAE,SAAS,EAAE,MAAM,SAAS,CAAA;AACnC,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAA;AAEnC,MAAM,SAAS,GAAG;;;;;;;;;;;;;;;;;;;;;;CAsBjB,CAAA;AAmBD,MAAM,OAAO,mBAAmB;IAK9B,YACU,QAAmB,EAC3B,OAA4B,EAAE;QADtB,aAAQ,GAAR,QAAQ,CAAW;QAJrB,QAAG,GAAG,CAAC,CAAA;QACP,eAAU,GAAsB,IAAI,CAAA;QAM1C,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,IAAI,qBAAqB,CAAA;QACnD,SAAS,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;QAE/C,IAAI,CAAC,EAAE,GAAG,IAAI,QAAQ,CAAC,MAAM,CAAC,CAAA;QAC9B,IAAI,CAAC,EAAE,CAAC,MAAM,CAAC,oBAAoB,CAAC,CAAA;QACpC,IAAI,CAAC,EAAE,CAAC,MAAM,CAAC,qBAAqB,CAAC,CAAA;QACrC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;QAEvB,0BAA0B;QAC1B,MAAM,MAAM,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAC5B,mEAAmE,CACpE,CAAC,GAAG,EAAgC,CAAA;QACrC,IAAI,MAAM,EAAE,CAAC;YACX,MAAM,KAAK,GAAG,MAAM,CAAC,EAAE,CAAC,KAAK,CAAC,GAAG,CAAC,CAAA;YAClC,IAAI,CAAC,GAAG,GAAG,QAAQ,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,EAAE,EAAE,CAAC,IAAI,CAAC,CAAA;QACvD,CAAC;IACH,CAAC;IAED,KAAK,CAAC,GAAG,CACP,KAA6E;QAE7E,MAAM,KAAK,GAAmB;YAC5B,GAAG,KAAK;YACR,EAAE,EAAE,IAAI,CAAC,UAAU,EAAE;YACrB,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE;YACrB,WAAW,EAAE,CAAC;YACd,SAAS,EAAE,GAAG;SACf,CAAA;QAED,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC;;;;;;KAMf,CAAC,CAAC,GAAG,CACJ,KAAK,CAAC,EAAE,EACR,KAAK,CAAC,IAAI,EACV,KAAK,CAAC,KAAK,EACX,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,IAAI,CAAC,EAC1B,KAAK,CAAC,UAAU,EAChB,KAAK,CAAC,WAAW,IAAI,IAAI,EACzB,KAAK,CAAC,SAAS,EACf,KAAK,CAAC,WAAW,EACjB,KAAK,CAAC,YAAY,IAAI,IAAI,EAC1B,KAAK,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EACtB,KAAK,CAAC,UAAU,IAAI,IAAI,EACxB,KAAK,CAAC,UAAU,IAAI,IAAI,EACxB,KAAK,CAAC,SAAS,EACf,KAAK,CAAC,cAAc,IAAI,IAAI,CAC7B,CAAA;QAED,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,iBAAiB,EAAE,EAAE,QAAQ,EAAE,KAAK,CAAC,EAAE,EAAE,EAAE,EAAE,UAAU,EAAE,KAAK,CAAC,cAAc,EAAE,CAAC,CAAA;QAEnG,sBAAsB;QACtB,IAAI,IAAI,CAAC,UAAU,EAAE,CAAC;YACpB,MAAM,IAAI,GAAG,CAAC,KAAK,CAAC,KAAK,EAAE,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,KAAK,CAAC,UAAU,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;YAClG,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,EAAE,IAAI,CAAC,CAAA;QACxC,CAAC;QAED,OAAO,KAAK,CAAA;IACd,CAAC;IAED,KAAK,CAAC,MAAM,CAAC,KAAqB;QAChC,MAAM,UAAU,GAAa,CAAC,KAAK,CAAC,CAAA;QACpC,MAAM,MAAM,GAAc,EAAE,CAAA;QAE5B,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC;YACf,MAAM,KAAK,GAAG,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,CAAA;YACnE,UAAU,CAAC,IAAI,CAAC,YAAY,KAAK,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;YAC9D,MAAM,CAAC,IAAI,CAAC,GAAG,KAAK,CAAC,CAAA;QACvB,CAAC;QAED,IAAI,KAAK,CAAC,IAAI,IAAI,KAAK,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACxC,KAAK,MAAM,GAAG,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC;gBAC7B,UAAU,CAAC,IAAI,CAAC,aAAa,CAAC,CAAA;gBAC9B,MAAM,CAAC,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,CAAA;YAC3B,CAAC;QACH,CAAC;QAED,IAAI,KAAK,CAAC,YAAY,KAAK,SAAS,EAAE,CAAC;YACrC,UAAU,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YACjC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,CAAA;QACjC,CAAC;QAED,IAAI,KAAK,CAAC,YAAY,EAAE,CAAC;YACvB,UAAU,CAAC,IAAI,CAAC,cAAc,CAAC,CAAA;QACjC,CAAC;QAED,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,IAAI,EAAE,CAAA;QAC/B,MAAM,GAAG,GAAG,yCAAyC,UAAU,CAAC,IAAI,CAAC,OAAO,CAAC,kCAAkC,CAAA;QAC/G,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QAElB,MAAM,IAAI,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,GAAG,MAAM,CAAY,CAAA;QAC3D,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAA;IACzC,CAAC;IAED,KAAK,CAAC,cAAc,CAAC,IAAY,EAAE,IAAY;QAC7C,0CAA0C;QAC1C,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,CAAC;YACrB,IAAI,CAAC,UAAU,GAAG,IAAI,UAAU,EAAE,CAAA;YAClC,MAAM,IAAI,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,4DAA4D,CAAC,CAAC,GAAG,EAAa,CAAA;YAC3G,IAAI,CAAC,UAAU,CAAC,aAAa,CAAC,IAAI,CAAC,CAAA;YACnC,MAAM,CAAC,KAAK,CAAC,EAAE,QAAQ,EAAE,IAAI,CAAC,MAAM,EAAE,EAAE,oBAAoB,CAAC,CAAA;QAC/D,CAAC;QAED,MAAM,OAAO,GAAG,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;QAClD,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACzB,MAAM,CAAC,KAAK,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,kCAAkC,CAAC,CAAA;YAChE,OAAO,IAAI,CAAC,MAAM,CAAC,EAAE,YAAY,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAA;QACzD,CAAC;QAED,MAAM,OAAO,GAAqB,EAAE,CAAA;QACpC,KAAK,MAAM,EAAE,EAAE,EAAE,KAAK,EAAE,IAAI,OAAO,EAAE,CAAC;YACpC,MAAM,GAAG,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,8CAA8C,CAAC,CAAC,GAAG,CAAC,EAAE,CAAsB,CAAA;YACxG,IAAI,GAAG,EAAE,CAAC;gBACR,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAA;gBAC/B,sCAAsC;gBACtC,KAAK,CAAC,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,SAAS,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC,CAAA;gBAC5D,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;YACrB,CAAC;QACH,CAAC;QAED,MAAM,CAAC,KAAK,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,CAAC,MAAM,EAAE,EAAE,yBAAyB,CAAC,CAAA;QAChF,OAAO,OAAO,CAAA;IAChB,CAAC;IAED,KAAK,CAAC,WAAW,CAAC,EAAU,EAAE,SAAiB;QAC7C,MAAM,GAAG,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,8CAA8C,CAAC,CAAC,GAAG,CAAC,EAAE,CAAsB,CAAA;QACxG,IAAI,CAAC,GAAG;YAAE,OAAM;QAEhB,MAAM,YAAY,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,GAAG,CAAC,SAAS,GAAG,IAAI,CAAC,CAAA;QACtD,MAAM,QAAQ,GAAG,GAAG,CAAC,YAAY,GAAG,CAAC,CAAA;QACrC,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;QAEtB,IAAI,CAAC,EAAE,CAAC,OAAO,CACb,8FAA8F,CAC/F,CAAC,GAAG,CAAC,YAAY,EAAE,QAAQ,EAAE,GAAG,EAAE,EAAE,CAAC,CAAA;QAEtC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,gBAAgB,EAAE,EAAE,QAAQ,EAAE,EAAE,EAAE,EAAE,EAAE,SAAS,EAAE,CAAC,CAAA;IACvE,CAAC;IAED,KAAK,CAAC,WAAW,CAAC,EAAU,EAAE,SAAiB;QAC7C,MAAM,GAAG,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,8CAA8C,CAAC,CAAC,GAAG,CAAC,EAAE,CAAsB,CAAA;QACxG,IAAI,CAAC,GAAG;YAAE,OAAM;QAEhB,MAAM,YAAY,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,GAAG,CAAC,SAAS,GAAG,GAAG,CAAC,CAAA;QACxD,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,yDAAyD,CAAC,CAAC,GAAG,CAAC,YAAY,EAAE,EAAE,CAAC,CAAA;QAChG,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,gBAAgB,EAAE,EAAE,QAAQ,EAAE,EAAE,EAAE,EAAE,EAAE,SAAS,EAAE,CAAC,CAAA;IACvE,CAAC;IAED,KAAK,CAAC,MAAM,CAAC,EAAU,EAAE,UAAkB;QACzC,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;QACtB,MAAM,MAAM,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAC5B,0FAA0F,CAC3F,CAAC,GAAG,CAAC,UAAU,EAAE,GAAG,EAAE,EAAE,CAAC,CAAA;QAE1B,IAAI,MAAM,CAAC,OAAO,GAAG,CAAC,EAAE,CAAC;YACvB,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,iBAAiB,EAAE,EAAE,QAAQ,EAAE,EAAE,EAAE,UAAU,EAAE,CAAC,CAAA;QACrE,CAAC;IACH,CAAC;IAED,KAAK,CAAC,KAAK;QACT,MAAM,GAAG,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,IAAI,CAAA;QAC/B,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;QACtB,MAAM,IAAI,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,4DAA4D,CAAC,CAAC,GAAG,EAI5F,CAAA;QAEF,MAAM,IAAI,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,yDAAyD,CAAC,CAAA;QACvF,MAAM,SAAS,GAAG,IAAI,CAAC,EAAE,CAAC,WAAW,CAAC,GAAG,EAAE;YACzC,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;gBACvB,MAAM,IAAI,GAAG,GAAG,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,GAAG,CAAC,aAAa,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,EAAE,CAAA;gBACrE,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,GAAG,EAAE,CAAC,CAAA;gBACpC,MAAM,YAAY,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,GAAG,CAAC,SAAS,GAAG,IAAI,GAAG,OAAO,GAAG,IAAI,CAAC,CAAA;gBAC1E,IAAI,CAAC,GAAG,CAAC,YAAY,EAAE,GAAG,CAAC,EAAE,CAAC,CAAA;YAChC,CAAC;QACH,CAAC,CAAC,CAAA;QACF,SAAS,EAAE,CAAA;IACb,CAAC;IAED,gCAAgC;IAChC,KAAK;QACH,IAAI,CAAC,EAAE,CAAC,KAAK,EAAE,CAAA;IACjB,CAAC;IAED,gBAAgB;IAChB,KAAK;QACH,MAAM,UAAU,GAAI,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,6CAA6C,CAAC,CAAC,GAAG,EAAoB,CAAC,CAAC,CAAA;QAC5G,MAAM,aAAa,GAAI,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,gEAAgE,CAAC,CAAC,GAAG,EAAoB,CAAC,CAAC,CAAA;QAClI,MAAM,IAAI,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,iEAAiE,CAAC,CAAC,GAAG,EAAmC,CAAA;QACtI,MAAM,MAAM,GAA2B,EAAE,CAAA;QACzC,KAAK,MAAM,CAAC,IAAI,IAAI;YAAE,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAA;QAC1C,OAAO,EAAE,UAAU,EAAE,aAAa,EAAE,MAAM,EAAE,CAAA;IAC9C,CAAC;IAEO,UAAU;QAChB,IAAI,CAAC,GAAG,GAAG,CAAC,IAAI,CAAC,GAAG,GAAG,CAAC,CAAC,GAAG,KAAK,CAAA;QACjC,OAAO,MAAM,IAAI,CAAC,GAAG,EAAE,IAAI,IAAI,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,QAAQ,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,CAAA;IACnE,CAAC;IAEO,OAAO,CAAC,GAAU;QACxB,OAAO;YACL,EAAE,EAAE,GAAG,CAAC,EAAE;YACV,IAAI,EAAE,GAAG,CAAC,IAA8B;YACxC,KAAK,EAAE,GAAG,CAAC,KAAK;YAChB,IAAI,EAAE,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAa;YACtC,UAAU,EAAE,GAAG,CAAC,WAAW;YAC3B,WAAW,EAAE,GAAG,CAAC,YAAY,IAAI,SAAS;YAC1C,SAAS,EAAE,GAAG,CAAC,SAAS;YACxB,WAAW,EAAE,GAAG,CAAC,YAAY;YAC7B,YAAY,EAAE,GAAG,CAAC,aAAa,IAAI,SAAS;YAC5C,QAAQ,EAAE,GAAG,CAAC,QAAQ,KAAK,CAAC;YAC5B,UAAU,EAAE,GAAG,CAAC,WAAW,IAAI,SAAS;YACxC,UAAU,EAAE,GAAG,CAAC,WAAW,IAAI,SAAS;YACxC,SAAS,EAAE,GAAG,CAAC,UAAU;YACzB,cAAc,EAAE,GAAG,CAAC,eAAe,IAAI,SAAS;SACjD,CAAA;IACH,CAAC;CACF"}

package/dist/knowledge/TfidfIndex.d.ts

@@ -0,0 +1,50 @@
+export interface TfidfSearchResult {
+    id: string;
+    score: number;
+}
+export declare class TfidfIndex {
+    private docTerms;
+    private docFreq;
+    private docCount;
+    /**
+     * 添加/更新文档到索引
+     */
+    upsert(id: string, text: string): void;
+    /**
+     * 从索引中移除文档
+     */
+    remove(id: string): void;
+    /**
+     * 搜索最相似的 topK 个文档
+     */
+    search(query: string, topK: number): TfidfSearchResult[];
+    /**
+     * 从行数据批量构建索引
+     */
+    buildFromRows(rows: Array<{
+        id: string;
+        title: string;
+        tags: string;
+        content_ref: string;
+    }>): void;
+    /**
+     * 清空索引
+     */
+    clear(): void;
+    /**
+     * 获取索引大小
+     */
+    get size(): number;
+    /**
+     * 分词：小写 + 按非字母数字分割 + 过滤停用词和短词
+     */
+    private tokenize;
+    /**
+     * 计算 TF-IDF 向量
+     */
+    private tfidfVector;
+    /**
+     * 计算余弦相似度
+     */
+    private cosineSimilarity;
+}

package/dist/knowledge/TfidfIndex.js

@@ -0,0 +1,177 @@
+// SCALE Engine — TF-IDF Index
+// 本地向量搜索，零外部依赖，基于 TF-IDF + 余弦相似度
+export class TfidfIndex {
+    constructor() {
+        this.docTerms = new Map();
+        this.docFreq = new Map(); // term → 出现在多少文档中
+        this.docCount = 0;
+    }
+    /**
+     * 添加/更新文档到索引
+     */
+    upsert(id, text) {
+        // 如果已存在，先移除旧的
+        if (this.docTerms.has(id)) {
+            this.remove(id);
+        }
+        const terms = this.tokenize(text);
+        this.docTerms.set(id, terms);
+        this.docCount++;
+        // 更新文档频率
+        const uniqueTerms = new Set(terms);
+        for (const term of uniqueTerms) {
+            this.docFreq.set(term, (this.docFreq.get(term) ?? 0) + 1);
+        }
+    }
+    /**
+     * 从索引中移除文档
+     */
+    remove(id) {
+        const terms = this.docTerms.get(id);
+        if (!terms)
+            return;
+        const uniqueTerms = new Set(terms);
+        for (const term of uniqueTerms) {
+            const count = this.docFreq.get(term) ?? 0;
+            if (count <= 1) {
+                this.docFreq.delete(term);
+            }
+            else {
+                this.docFreq.set(term, count - 1);
+            }
+        }
+        this.docTerms.delete(id);
+        this.docCount--;
+    }
+    /**
+     * 搜索最相似的 topK 个文档
+     */
+    search(query, topK) {
+        if (this.docCount === 0)
+            return [];
+        const queryTerms = this.tokenize(query);
+        if (queryTerms.length === 0)
+            return [];
+        const queryVec = this.tfidfVector(queryTerms);
+        if (queryVec.size === 0)
+            return [];
+        const results = [];
+        for (const [id, docTerms] of this.docTerms) {
+            const docVec = this.tfidfVector(docTerms);
+            const score = this.cosineSimilarity(queryVec, docVec);
+            if (score > 0) {
+                results.push({ id, score });
+            }
+        }
+        // 按分数降序排序
+        results.sort((a, b) => b.score - a.score);
+        return results.slice(0, topK);
+    }
+    /**
+     * 从行数据批量构建索引
+     */
+    buildFromRows(rows) {
+        this.clear();
+        for (const row of rows) {
+            // 组合 title + tags + content_ref 作为搜索文本
+            const text = [row.title, row.tags, row.content_ref].filter(Boolean).join(' ');
+            this.upsert(row.id, text);
+        }
+    }
+    /**
+     * 清空索引
+     */
+    clear() {
+        this.docTerms.clear();
+        this.docFreq.clear();
+        this.docCount = 0;
+    }
+    /**
+     * 获取索引大小
+     */
+    get size() {
+        return this.docCount;
+    }
+    /**
+     * 分词：小写 + 按非字母数字分割 + 过滤停用词和短词
+     */
+    tokenize(text) {
+        const stopWords = new Set([
+            'the', 'a', 'an', 'is', 'are', 'was', 'were', 'be', 'been', 'being',
+            'have', 'has', 'had', 'do', 'does', 'did', 'will', 'would', 'could',
+            'should', 'may', 'might', 'shall', 'can', 'need', 'dare', 'ought',
+            'and', 'or', 'but', 'if', 'then', 'else', 'when', 'at', 'from',
+            'by', 'for', 'with', 'about', 'against', 'between', 'through',
+            'during', 'before', 'after', 'above', 'below', 'to', 'from',
+            'up', 'down', 'in', 'out', 'on', 'off', 'over', 'under',
+            'again', 'further', 'then', 'once', 'here', 'there', 'when',
+            'where', 'why', 'how', 'all', 'both', 'each', 'few', 'more',
+            'most', 'other', 'some', 'such', 'no', 'nor', 'not', 'only',
+            'own', 'same', 'so', 'than', 'too', 'very', 's', 't', 'just',
+            'don', 'now',
+            // 中文停用词
+            '的', '了', '在', '是', '我', '有', '和', '就', '不', '人', '都',
+            '一', '一个', '上', '也', '很', '到', '说', '要', '去', '你',
+            '会', '着', '没有', '看', '好', '自己', '这',
+        ]);
+        const rawTokens = text
+            .toLowerCase()
+            .split(/[^a-z0-9一-鿿]+/)
+            .filter(token => token && !stopWords.has(token));
+        const tokens = [];
+        for (const token of rawTokens) {
+            if (/[一-鿿]/.test(token)) {
+                // CJK: emit each character as a token (words aren't space-separated)
+                for (const ch of token) {
+                    if (/[一-鿿]/.test(ch))
+                        tokens.push(ch);
+                }
+            }
+            else if (token.length >= 2) {
+                tokens.push(token);
+            }
+        }
+        return tokens;
+    }
+    /**
+     * 计算 TF-IDF 向量
+     */
+    tfidfVector(terms) {
+        const tfidf = new Map();
+        const uniqueTerms = new Set(terms);
+        // 计算词频 (TF)
+        const tf = new Map();
+        for (const term of terms) {
+            tf.set(term, (tf.get(term) ?? 0) + 1);
+        }
+        for (const term of uniqueTerms) {
+            const termFreq = (tf.get(term) ?? 0) / terms.length;
+            const docFreq = this.docFreq.get(term) ?? 0;
+            const idf = docFreq > 0 ? Math.log(this.docCount / docFreq) : 0;
+            const weight = termFreq * idf;
+            if (weight > 0)
+                tfidf.set(term, weight);
+        }
+        return tfidf;
+    }
+    /**
+     * 计算余弦相似度
+     */
+    cosineSimilarity(a, b) {
+        if (a.size === 0 || b.size === 0)
+            return 0;
+        // 点积
+        let dot = 0;
+        for (const [term, weight] of a) {
+            if (b.has(term))
+                dot += weight * b.get(term);
+        }
+        // 模
+        const magA = Math.sqrt(Array.from(a.values()).reduce((s, w) => s + w * w, 0));
+        const magB = Math.sqrt(Array.from(b.values()).reduce((s, w) => s + w * w, 0));
+        if (magA === 0 || magB === 0)
+            return 0;
+        return dot / (magA * magB);
+    }
+}
+//# sourceMappingURL=TfidfIndex.js.map