feat(scoring): implemented semantic context padding and optimized excerpting v2.76.0

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "g1nation",
-  "version": "2.75.0",
+  "version": "2.76.0",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "g1nation",
-      "version": "2.75.0",
+      "version": "2.76.0",
       "license": "MIT",
       "dependencies": {
         "marked": "^18.0.2"

package.json

@@ -2,7 +2,7 @@
   "name": "astra",
   "displayName": "Astra",
   "description": "The personal intelligence layer for Antigravity and VS Code. A private cognitive partner for deep project context, memory, and proactive strategic decision-making.",
-  "version": "2.75.0",
+  "version": "2.76.0",
   "publisher": "g1nation",
   "license": "MIT",
   "icon": "assets/icon.png",

src/retrieval/scoring.ts

@@ -300,7 +300,7 @@ export function extractBestExcerpt(
     const sentences = content
         .split(/(?<=[.!?。！？\n])\s*/)
         .map((s) => s.trim())
-        .filter((s) => s.length > 10);
+        .filter((s) => s.length > 5);
 
     if (sentences.length === 0) return content.slice(0, maxLength);
 
@@ -343,8 +343,16 @@ export function extractBestExcerpt(
         }
     }
 
+    // 4. Result construction with semantic context padding
+    let finalStart = bestStart;
+    let finalEnd = bestStart + bestLen;
+
+    // 전후 문맥을 1문장씩 추가하여 의미적 완전성 확보 (예산 허용 시)
+    if (finalStart > 0) finalStart--;
+    if (finalEnd < scored.length) finalEnd++;
+
     const excerptSentences = scored
-        .slice(bestStart, bestStart + bestLen)
+        .slice(finalStart, finalEnd)
         .map((s) => s.sentence);
     
     const result = excerptSentences.join(' ');

tests/scoring.test.ts

@@ -94,6 +94,21 @@ describe('Scoring Engine Unit Tests (v2.72.0)', () => {
         expect(duration).toBeLessThan(100); // Tokenizer should be efficient even for long text
     });
 
+    test('Contextual Completeness: should include adjacent sentences for semantic padding', () => {
+        const content = `
+            도입부 문장입니다.
+            핵심 키워드 성능 최적화가 포함된 문장입니다.
+            마무리 문장입니다.
+        `;
+        const query = ['성능', '최적화'];
+        const excerpt = extractBestExcerpt(content, query, 200);
+        
+        // Should include introduction and conclusion due to padding
+        expect(excerpt).toContain('도입부 문장');
+        expect(excerpt).toContain('핵심 키워드');
+        expect(excerpt).toContain('마무리 문장');
+    });
+
     test('Performance Benchmark: should process 100 documents within threshold', () => {
         const query = tokenize('performance optimization');
         const largeDocs = Array.from({ length: 100 }, (_, i) => ({