From eb64556636bd2bee50ea4bb12704f1f20bed6e39 Mon Sep 17 00:00:00 2001
From: matevip <matevip@mate.vip>
Date: Mon, 4 May 2026 20:54:44 +0800
Subject: [PATCH] fix(wiki): skip per-turn retrieval for continuation messages
 + drop weak tail hits

---
 .../java/vip/mate/wiki/WikiProperties.java    | 16 +++++++++
 .../mate/wiki/service/WikiContextService.java | 36 +++++++++++++++++++
 2 files changed, 52 insertions(+)

diff --git a/mateclaw-server/src/main/java/vip/mate/wiki/WikiProperties.java b/mateclaw-server/src/main/java/vip/mate/wiki/WikiProperties.java
index 8a3df9f1..cf3b4c00 100644
--- a/mateclaw-server/src/main/java/vip/mate/wiki/WikiProperties.java
+++ b/mateclaw-server/src/main/java/vip/mate/wiki/WikiProperties.java
@@ -133,6 +133,22 @@ public class WikiProperties {
     /** 混合搜索默认模式：keyword / semantic / hybrid */
     private String searchDefaultMode = "hybrid";
 
+    /**
+     * Minimum trimmed char length for a user message before per-turn wiki
+     * retrieval kicks in. Short messages like "继续" / "嗯" / "OK" carry no
+     * semantic signal and the retriever falls back to whichever pages happen
+     * to dominate the index, polluting the prompt with off-topic content.
+     */
+    private int relevantContextMinQueryLength = 3;
+
+    /**
+     * Relative score floor for relevant-wiki hits. A hit is dropped when its
+     * score is below {@code topHit.score * this ratio}, so a single strong
+     * match does not drag in low-relevance tail pages alongside it. Set to
+     * 0 to disable.
+     */
+    private double relevantContextMinRelativeScore = 0.5;
+
     // ==================== RFC-031: Light processing tiers ====================
 
     /** Whether to auto-dispatch a LIGHT_ENRICH job after heavy ingest completes */
diff --git a/mateclaw-server/src/main/java/vip/mate/wiki/service/WikiContextService.java b/mateclaw-server/src/main/java/vip/mate/wiki/service/WikiContextService.java
index e366276c..d40c2fa1 100644
--- a/mateclaw-server/src/main/java/vip/mate/wiki/service/WikiContextService.java
+++ b/mateclaw-server/src/main/java/vip/mate/wiki/service/WikiContextService.java
@@ -9,6 +9,8 @@ import vip.mate.wiki.model.WikiKnowledgeBaseEntity;
 import vip.mate.wiki.model.WikiPageEntity;
 
 import java.util.List;
+import java.util.Locale;
+import java.util.Set;
 
 /**
  * Wiki context service — builds context for agent conversation injection.
@@ -26,6 +28,18 @@ public class WikiContextService {
     private final HybridRetriever hybridRetriever;
     private final WikiProperties properties;
 
+    /**
+     * Continuation / acknowledgement tokens. When the user message reduces to
+     * one of these the per-turn retrieval is skipped — these queries carry no
+     * topical signal and the retriever otherwise returns whichever pages
+     * dominate the index, derailing the conversation onto unrelated content.
+     */
+    private static final Set<String> CONTINUATION_TOKENS = Set.of(
+            "继续", "继续吧", "继续做", "继续输出", "继续完成", "请继续",
+            "好的", "好", "嗯", "嗯嗯", "是的", "对", "对的",
+            "ok", "okay", "go on", "next", "continue", "yes", "yeah", "sure", "y"
+    );
+
     /**
      * Build relevant wiki context for the current user message.
      * <p>
@@ -37,6 +51,16 @@ public class WikiContextService {
             return "";
         }
 
+        // Skip retrieval for continuation / acknowledgement turns — observed
+        // in production: a user reply of "继续" produced top-5 hits dominated
+        // by the wiki KB's most-cited pages and rerouted four parallel
+        // multi-agent tasks onto unrelated topics.
+        String trimmed = userMessage.trim();
+        if (trimmed.length() < properties.getRelevantContextMinQueryLength()
+                || CONTINUATION_TOKENS.contains(trimmed.toLowerCase(Locale.ROOT))) {
+            return "";
+        }
+
         List<WikiKnowledgeBaseEntity> kbs = kbService.listByAgentId(agentId);
         if (kbs.isEmpty()) {
             return "";
@@ -48,6 +72,18 @@ public class WikiContextService {
             return "";
         }
 
+        // Drop tail hits that score far below the top hit so a single strong
+        // match isn't accompanied by 4 weakly-related pages.
+        double minRel = properties.getRelevantContextMinRelativeScore();
+        if (minRel > 0 && hits.size() > 1) {
+            double topScore = hits.get(0).score();
+            double floor = topScore * minRel;
+            hits = hits.stream().filter(h -> h.score() >= floor).toList();
+            if (hits.isEmpty()) {
+                return "";
+            }
+        }
+
         StringBuilder sb = new StringBuilder("<wiki-relevant>\n");
         sb.append("[Relevant wiki pages for this query. Use wiki_read_page(slug) for full content. " +
                 "When using information from these pages in your answer, always cite the source page title, " +