@vespermcp/mcp-server 1.2.25 → 1.2.26

package/build/web/fusion-engine.js

@@ -48,14 +48,45 @@ function titleTokens(doc) {
     const raw = typeof mj.title === "string" ? mj.title : "";
     return tokenize(raw);
 }
+function semanticHintTokens(doc) {
+    const mj = doc.metadata_json || {};
+    const fields = [];
+    if (typeof mj.title === "string")
+        fields.push(mj.title);
+    if (typeof mj.name === "string")
+        fields.push(mj.name);
+    if (typeof mj.description === "string")
+        fields.push(mj.description);
+    if (typeof mj.abstract === "string")
+        fields.push(mj.abstract);
+    if (Array.isArray(mj.tags))
+        fields.push(mj.tags.join(" "));
+    if (Array.isArray(mj.topics))
+        fields.push(mj.topics.join(" "));
+    fields.push(doc.source_url || "");
+    return tokenize(fields.join(" "));
+}
 function isSuspiciousPair(a, b) {
     // semantic fallback should be selective; do cheap prefilter first
+    // Metadata/topic overlap can indicate same object even with very different body lengths.
+    const aHints = semanticHintTokens(a);
+    const bHints = semanticHintTokens(b);
+    if (aHints.size > 0 && bHints.size > 0) {
+        let hInter = 0;
+        for (const t of aHints)
+            if (bHints.has(t))
+                hInter++;
+        const hUnion = aHints.size + bHints.size - hInter;
+        const hJaccard = hUnion > 0 ? hInter / hUnion : 0;
+        if (hJaccard >= 0.2)
+            return true;
+    }
     const aLen = a.content.length;
     const bLen = b.content.length;
     const maxLen = Math.max(aLen, bLen, 1);
     const lenRatio = Math.abs(aLen - bLen) / maxLen;
     // Loosened again to allow abstract-vs-summary style comparisons.
-    if (lenRatio > 0.8)
+    if (lenRatio > 0.9)
         return false;
     // Fast path: same normalized title-like prefix often indicates same research object.
     const aPrefix = a.content.slice(0, 140).toLowerCase().replace(/[^a-z0-9\s]/g, " ").trim();

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@vespermcp/mcp-server",
-  "version": "1.2.25",
+  "version": "1.2.26",
   "description": "AI-powered dataset discovery, quality analysis, and preparation MCP server with multimodal support (text, image, audio, video)",
   "type": "module",
   "main": "build/index.js",