fix(aihr): trim sop evidence snippets

2026-07-09 15:37:01 +08:00
parent ba8c35904e
commit 2ef3e39744
2 changed files with 72 additions and 5 deletions
@@ -97,6 +97,7 @@ public class AihrSopSeedService {
    private static final String LOCAL_EMBEDDING_MODEL = "local-hash-v1";
    private static final String LOCAL_EMBEDDING_BASE_URL = "local";
    private static final int LOCAL_EMBEDDING_DIMENSION = 1536;
+    private static final int MAX_EVIDENCE_SNIPPETS = 3;
    private static final String NO_CLEAR_SOP_EVIDENCE = "未在已入库 SOP 中找到明确依据";
    private static final String SOP_ANSWER_PROMPT_VERSION = "builtin-sop_answer-v1";
    private static final String SOP_ANSWER_SYSTEM_PROMPT = "你是物业一线员工的SOP助手。只返回 JSON，不要 markdown。字段固定为 answer, keyPoints, cautions。"
@@ -2209,16 +2210,17 @@ public class AihrSopSeedService {
    }

    private SearchResponse responseFromHits(String category, String queryText, List<KnowledgeHit> hits, List<String> extraCautions) {
-        KnowledgeHit first = hits.get(0);
+        List<KnowledgeHit> evidenceHits = evidenceHits(queryText, hits);
+        KnowledgeHit first = evidenceHits.get(0);
        String responseCategory = isBlank(category) ? first.category() : category;
-        List<SnippetResponse> snippets = hits.stream()
+        List<SnippetResponse> snippets = evidenceHits.stream()
            .map(hit -> new SnippetResponse("第 " + hit.idx() + " 段：" + hit.title(), hit.content(), hit.fragmentId()))
            .toList();
        Map<String, DocResponse> docs = new LinkedHashMap<>();
-        for (KnowledgeHit hit : hits) {
+        for (KnowledgeHit hit : evidenceHits) {
            docs.putIfAbsent(hit.title(), new DocResponse(hit.title(), hit.description(), "命中 " + scoreText(hit.score()), "pill-success"));
        }
-        SearchDigest digest = searchDigest(queryText, hits);
+        SearchDigest digest = searchDigest(queryText, evidenceHits);
        if (digest == null) {
            return rawHitResponse(responseCategory, queryText, List.copyOf(docs.values()), snippets, extraCautions);
        }
@@ -2240,6 +2242,36 @@ public class AihrSopSeedService {
        );
    }

+    static List<KnowledgeHit> evidenceHits(String queryText, List<KnowledgeHit> hits) {
+        if (hits == null || hits.isEmpty()) {
+            return List.of();
+        }
+        KnowledgeHit first = hits.get(0);
+        if (queryMentionsTitle(queryText, first.title())) {
+            List<KnowledgeHit> sameTitle = hits.stream()
+                .filter(hit -> Objects.equals(hit.title(), first.title()))
+                .limit(MAX_EVIDENCE_SNIPPETS)
+                .toList();
+            if (!sameTitle.isEmpty()) {
+                return sameTitle;
+            }
+        }
+        return hits.stream().limit(MAX_EVIDENCE_SNIPPETS).toList();
+    }
+
+    private static boolean queryMentionsTitle(String queryText, String title) {
+        String query = evidenceText(queryText);
+        String docTitle = evidenceText(title);
+        return query.length() >= 4 && docTitle.length() >= 4 && (query.contains(docTitle) || docTitle.contains(query));
+    }
+
+    private static String evidenceText(String value) {
+        return Optional.ofNullable(value).orElse("")
+            .toLowerCase(Locale.ROOT)
+            .replaceFirst("\\.[a-z0-9]{1,8}$", "")
+            .replaceAll("[^\\p{IsHan}\\p{Alnum}]+", "");
+    }
+
    static SearchResponse rawHitResponse(String category, String queryText, List<DocResponse> docs, List<SnippetResponse> snippets, List<String> extraCautions) {
        if (snippets == null || snippets.isEmpty()) {
            return noEvidenceResponse(queryText, category, extraCautions);
@@ -3283,7 +3315,7 @@ public class AihrSopSeedService {
    ) {
    }

-    private record KnowledgeHit(Long fragmentId, String title, String category, String description, String docId, String content, int idx, double score) {
+    record KnowledgeHit(Long fragmentId, String title, String category, String description, String docId, String content, int idx, double score) {
    }

    private record FeedbackFilterResult(List<KnowledgeHit> hits, List<String> cautions) {
@@ -33,4 +33,39 @@ public class AihrSopSeedServiceTest {
        assertTrue(AihrSopSeedService.skipSopReview("verify_demo_questions"));
        assertFalse(AihrSopSeedService.skipSopReview("knowledge_search"));
    }
+
+    @Test
+    @Tag("dev")
+    public void evidenceHitsKeepsNamedDocumentOnly() {
+        List<AihrSopSeedService.KnowledgeHit> hits = List.of(
+            hit(1L, "生活顾问的一天.mp4"),
+            hit(2L, "生活顾问的一天.mp4"),
+            hit(3L, "信息安全管理培训.pptx")
+        );
+
+        List<AihrSopSeedService.KnowledgeHit> evidence = AihrSopSeedService.evidenceHits("生活顾问的一天做什么", hits);
+
+        assertEquals(2, evidence.size());
+        assertTrue(evidence.stream().allMatch(hit -> "生活顾问的一天.mp4".equals(hit.title())));
+    }
+
+    @Test
+    @Tag("dev")
+    public void evidenceHitsCapsGenericQueries() {
+        List<AihrSopSeedService.KnowledgeHit> hits = List.of(
+            hit(1L, "投诉处理 SOP.pdf"),
+            hit(2L, "催缴沟通 SOP.pdf"),
+            hit(3L, "报修跟进 SOP.pdf"),
+            hit(4L, "保洁合同.pdf")
+        );
+
+        List<AihrSopSeedService.KnowledgeHit> evidence = AihrSopSeedService.evidenceHits("生活顾问怎么处理业主问题", hits);
+
+        assertEquals(3, evidence.size());
+        assertEquals(3L, evidence.get(2).fragmentId());
+    }
+
+    private static AihrSopSeedService.KnowledgeHit hit(Long fragmentId, String title) {
+        return new AihrSopSeedService.KnowledgeHit(fragmentId, title, "sop", "", "doc-" + fragmentId, "片段内容", 1, 1.0);
+    }
 }