fix(search): prioritize titles and author/guest names over transcripts

Author: codercatdev

apps/site/src/lib/sanity/queries/search.ts

@@ -2,21 +2,25 @@ import { defineQuery } from "groq";
 
 /**
  * Hybrid semantic search utilizing Sanity dataset embeddings (text::semanticSimilarity)
- * combined with boost() scoring on title, excerpt, and transcript content.
+ * combined with heavy boost() scoring on title, name, and excerpt to guarantee
+ * direct title/name matches strictly outrank long transcript matches.
  */
 export const semanticSearchQuery = defineQuery(`
   *[_type in ["post", "podcast", "author", "guest"] && defined(slug.current) && ($type == null || $type == "all" || _type == $type)]
   | score(
-      boost(title match $searchTerm + "*", 5),
-      boost(excerpt match $searchTerm + "*", 2),
+      boost(lower(coalesce(title, name)) == lower($searchTerm), 100),
+      boost(coalesce(title, name) match $searchTerm + "*", 50),
+      boost(coalesce(title, name) match $searchTerm, 30),
+      boost(_type in ["author", "guest"] && coalesce(name, title) match $searchTerm + "*", 20),
+      boost(excerpt match $searchTerm + "*", 8),
       boost(transcript->summary match $searchTerm + "*", 3),
       boost(transcript->fullText match $searchTerm + "*", 1),
       text::semanticSimilarity($searchTerm)
     )
   | order(_score desc)[0...24] {
     _id,
     _type,
-    title,
+    "title": coalesce(title, name, "Untitled"),
     "slug": slug.current,
     "excerpt": coalesce(excerpt, transcript->summary),
     "transcriptSummary": transcript->summary,
@@ -31,21 +35,25 @@ export const semanticSearchQuery = defineQuery(`
  */
 export const textSearchFallbackQuery = defineQuery(`
   *[_type in ["post", "podcast", "author", "guest"] && defined(slug.current) && ($type == null || $type == "all" || _type == $type) && (
-    title match $searchTerm + "*" ||
+    coalesce(title, name) match $searchTerm + "*" ||
+    coalesce(title, name) match $searchTerm ||
     excerpt match $searchTerm + "*" ||
     transcript->summary match $searchTerm + "*" ||
     transcript->fullText match $searchTerm + "*"
   )]
   | score(
-      boost(title match $searchTerm + "*", 5),
-      boost(excerpt match $searchTerm + "*", 2),
+      boost(lower(coalesce(title, name)) == lower($searchTerm), 100),
+      boost(coalesce(title, name) match $searchTerm + "*", 50),
+      boost(coalesce(title, name) match $searchTerm, 30),
+      boost(_type in ["author", "guest"] && coalesce(name, title) match $searchTerm + "*", 20),
+      boost(excerpt match $searchTerm + "*", 8),
       boost(transcript->summary match $searchTerm + "*", 3),
       boost(transcript->fullText match $searchTerm + "*", 1)
     )
   | order(_score desc)[0...24] {
     _id,
     _type,
-    title,
+    "title": coalesce(title, name, "Untitled"),
     "slug": slug.current,
     "excerpt": coalesce(excerpt, transcript->summary),
     "transcriptSummary": transcript->summary,