@astermind/cybernetic-chatbot-client 2.2.18 → 2.2.30

package/dist/cybernetic-chatbot-client.umd.js

@@ -643,6 +643,10 @@
             // Tokenize and build document frequency
             const docFreq = new Map();
             for (const doc of documents) {
+                // Skip documents without content
+                if (!doc.content) {
+                    continue;
+                }
                 const tokens = this.tokenize(doc.content);
                 const uniqueTokens = new Set(tokens);
                 // Count document frequency
@@ -793,6 +797,9 @@
          * Tokenize text into words
          */
         tokenize(text) {
+            if (!text || typeof text !== 'string') {
+                return [];
+            }
             return text
                 .toLowerCase()
                 .replace(/[^\w\s]/g, ' ')
@@ -3546,22 +3553,35 @@ LJ5AZXvOhHaXdHzMuYKX5BpK4w7TqbPvJ6QPvKmLKvHh1VKcUJ6mJQgJJw==
             }
             return '';
         }
+        /**
+         * Strip common articles and filler words for better matching
+         */
+        stripArticles(text) {
+            return text
+                .replace(/^(the|a|an|my|our|your|this|that)\s+/gi, '')
+                .replace(/\s+(page|section|screen|view)$/gi, '')
+                .trim();
+        }
         /**
          * Find site map match for target string
          */
         findSiteMapMatch(target) {
             const normalizedTarget = target.toLowerCase().replace(/[^a-z0-9\s]/g, '');
-            // Exact match
-            const exact = this.siteMapIndex.get(normalizedTarget);
+            const strippedTarget = this.stripArticles(normalizedTarget);
+            // Exact match (try both with and without articles)
+            const exact = this.siteMapIndex.get(normalizedTarget) || this.siteMapIndex.get(strippedTarget);
             if (exact) {
                 return { entry: exact, similarity: 1.0 };
             }
-            // Fuzzy match
+            // Fuzzy match using stripped version for better accuracy
             let bestMatch = null;
             let bestScore = 0;
             for (const [key, entry] of this.siteMapIndex) {
-                const score = this.calculateSimilarity(normalizedTarget, key);
-                if (score > bestScore && score > 0.6) {
+                // Compare both original and stripped versions, take the better score
+                const score1 = this.calculateSimilarity(normalizedTarget, key);
+                const score2 = this.calculateSimilarity(strippedTarget, key);
+                const score = Math.max(score1, score2);
+                if (score > bestScore && score > 0.5) {
                     bestScore = score;
                     bestMatch = entry;
                 }