fix(knowledge): focus formal citation snippets

2026-07-31 01:17:15 +08:00
parent ca61b1f36f
commit 72c56e7c96
4 changed files with 64 additions and 7 deletions
@@ -74,12 +74,50 @@ public final class AihrFormalPolicyClassifier {
        if (query.isEmpty() || evidence.isEmpty()) {
            return false;
        }
-        for (EvidenceProfile profile : EVIDENCE_PROFILES) {
-            if (containsAny(query, profile.questionSignals())) {
-                return containsAny(evidence, profile.evidenceSignals());
-            }
+        EvidenceProfile profile = matchingProfile(query);
+        return profile == null || containsAny(evidence, profile.evidenceSignals());
+    }
+
+    /**
+     * Keep the visible citation centered on the same deterministic anchor used
+     * by the formal-evidence gate instead of returning an unrelated document prefix.
+     */
+    public static String relevantEvidenceSnippet(String question, String evidenceText, int maxLength) {
+        String evidence = evidenceText == null ? "" : evidenceText.trim();
+        if (evidence.isEmpty() || maxLength <= 0) {
+            return "";
        }
-        return true;
+        if (evidence.length() <= maxLength) {
+            return evidence;
+        }
+        EvidenceProfile profile = matchingProfile(question == null ? "" : question.trim());
+        if (profile == null || maxLength < 3) {
+            return evidence.substring(0, maxLength);
+        }
+        int anchorIndex = profile.evidenceSignals().stream()
+            .mapToInt(evidence::indexOf)
+            .filter(index -> index >= 0)
+            .min()
+            .orElse(-1);
+        if (anchorIndex < 0) {
+            return evidence.substring(0, maxLength);
+        }
+        int windowLength = maxLength - 2;
+        int start = Math.max(0, anchorIndex - Math.min(120, windowLength / 3));
+        int end = Math.min(evidence.length(), start + windowLength);
+        if (end == evidence.length()) {
+            start = Math.max(0, end - windowLength);
+        }
+        return (start > 0 ? "…" : "")
+            + evidence.substring(start, end).trim()
+            + (end < evidence.length() ? "…" : "");
+    }
+
+    private static EvidenceProfile matchingProfile(String question) {
+        return EVIDENCE_PROFILES.stream()
+            .filter(profile -> containsAny(question, profile.questionSignals()))
+            .findFirst()
+            .orElse(null);
    }

    private static EvidenceProfile evidence(List<String> questionSignals, List<String> evidenceSignals) {
@@ -650,7 +650,10 @@ public class AihrKnowledgeQueryService {
                || AihrFormalPolicyClassifier.isRelevantFormalEvidence(queryText, row.snippet()))
            .map(row -> new Citation(
                row.spaceCode(), row.sourceType(), row.docId(), row.title(),
-                truncate(row.snippet(), 500), row.fragmentId()
+                formalPolicyOnly
+                    ? AihrFormalPolicyClassifier.relevantEvidenceSnippet(queryText, row.snippet(), 500)
+                    : truncate(row.snippet(), 500),
+                row.fragmentId()
            ))
            .toList();
    }
@@ -72,4 +72,20 @@ class AihrFormalPolicyClassifierTest {
        assertFalse(AihrFormalPolicyClassifier.isRelevantFormalEvidence(
            installmentQuestion, " "));
    }
+
+    @Test
+    void formalCitationSnippetKeepsTheMatchedAnchorVisible() {
+        String question = "装修产生的废料该找谁、怎么约清运、能不能收费？";
+        String evidence = "无关前言。".repeat(120)
+            + "装修垃圾应登记后预约清运，收费标准须公示。"
+            + "无关附录。".repeat(120);
+
+        String snippet = AihrFormalPolicyClassifier.relevantEvidenceSnippet(question, evidence, 500);
+
+        assertTrue(snippet.length() <= 500);
+        assertTrue(snippet.contains("装修垃圾"));
+        assertTrue(snippet.contains("清运"));
+        assertTrue(snippet.startsWith("…"));
+        assertTrue(snippet.endsWith("…"));
+    }
 }