Tutorials for Vector Search and HFE

Author: ViktarStarastsenka

src/manifest.json

@@ -34,6 +34,26 @@
             "initialIsOpen": true,
             "path": "/uc/rag.md"
           }
+        },
+        {
+          "type": "internal-link",
+          "id": "ai_assistant",
+          "label": "Creating an AI Assistant",
+          "summary": "An assistant that temporarily retains conversations and understands by their meaning.",
+          "args": {
+            "initialIsOpen": true,
+            "path": "/uc/ai_assistant.md"
+          }
+        },
+        {
+          "type": "internal-link",
+          "id": "redis_use_cases_rag",
+          "label": "Building personalized recommendations",
+          "summary": "Not just keyword matching, but semantic understanding.",
+          "args": {
+            "initialIsOpen": true,
+            "path": "/uc/personalized_recommendations.md"
+          }
         }
       ]
     },

src/uc/ai_assistant.md

@@ -0,0 +1,94 @@
+Imagine you are building a smart AI assistant that:
+    - Remembers chats, but only temporarily
+    - Thinks by meaning, not by matching exact words
+    - Cleans up after itself, with no manual scripts
+
+Redis 8 has two powerful capabilities to make this happen:
+    - Field-level expiration: Let individual chat messages expire on their own
+    - Vector similarity search: Find past messages based on meaning, not keywords
+
+Let’s dive in.
+
+### Short-Term Memory with Field-Level Expiry
+Each chat session is stored as a Redis Hash.
+Each message is a field in the hash.
+Redis 8’s new HSETEX command allows you to assign a TTL to each field, perfect for building ephemeral, session-level memory.
+
+```redis:[run_confirmation=true] Upload Session Data
+// session:42 is the session ID
+// msg:<timestamp> ensures uniqueness and traceability
+HSETEX session:42 msg:1717935301 120 "Hi Chatbot!"
+HSETEX session:42 msg:1717935361 180 "What can you do?"
+HSETEX session:42 msg:1717935440 90 "Can you remind me about my tasks?"
+HSETEX session:42 msg:1717935720 30  "What's the news today?"
+```
+
+Each field automatically expires after its TTL (in seconds).
+No need for cron jobs or background workers.
+What you get:
+    - Clean memory
+    - Zero manual cleanup
+    - Session-scoped retention, just like short-term memory in humans
+
+
+Try it: After a few minutes, run `HGETALL session:42` and see what's left.
+
+### Vector Search for Semantic Recall
+Now, your assistant needs to “recall” semantically related messages, not just match by words.
+To do that, you’ll:
+    - Convert messages to vector embeddings
+    - Store them in Redis
+    - Use Vector Search with FT.SEARCH for semantic retrieval
+
+```redis:[run_confirmation=true] Create a Vector Index
+FT.CREATE idx:memory ON HASH PREFIX 1 memory: SCHEMA
+    message TEXT
+    embedding VECTOR FLAT // FLAT = exact vector search
+    6
+        TYPE FLOAT32
+        DIM 8 // DIM = embedding size, DIM 8 is just for demo purposes. In real use, embeddings are usually 128–1536 dimensions.
+        DISTANCE_METRIC COSINE // COSINE = measures semantic closeness
+```
+
+Now, let’s add entries for your chatbots:
+
+```redis:[run_confirmation=true] Add entries for the chatbot
+// Embeddings are stored as binary FLOAT32 vectors - this is a compact format required by Redis Vector Serch indexes
+HSET memory:1 message "Book a dentist appointment"       embedding "\x00\x00\x80?\x00\x00\x00@\x00\x00@@\x00\x00\x80@\x00\x00\x00@\x00\x00\x00@"
+HSET memory:2 message "Remind me to water plants"        embedding "\x00\x00\x80@\x00\x00\x80@\x00\x00\x80@\x00\x00\x80?\x00\x00\x80?\x00\x00@@"
+HSET memory:3 message "What’s the weather like?"         embedding "\x00\x00@@\x00\x00\x00@\x00\x00\x00@\x00\x00\x00@\x00\x00\x80?\x00\x00\x80?"
+HSET memory:4 message "Cancel my gym session"            embedding "\x00\x00@@\x00\x00\x00@\x00\x00\x80?\x00\x00\x80@\x00\x00\x00@\x00\x00\x00@"
+HSET memory:5 message "Start a new shopping list"        embedding "\x00\x00\x00@\x00\x00\x00@\x00\x00\x80?\x00\x00\x80@\x00\x00\x80?\x00\x00@@"
+```
+
+Now your messages are vectorized and ready for search.
+
+### Let Chatbot Think – Semantic Search with Vectors
+When a user sends a new message, convert it to an embedding and run a KNN search:
+
+```redis:[run_confirmation=true] Search For Similar Messages
+// Returns the top 3 semantically similar messages, even if no words match directly.
+FT.SEARCH idx:memory "*=>[KNN 3 @embedding $vec AS score]"
+   PARAMS 2 vec "\x00\x00@@\x00\x00\x80@\x00\x00\x00@\x00\x00\x80?\x00\x00@@\x00\x00\x00@"
+   SORTBY score
+   DIALECT 2
+```
+
+Now your assistant “remembers” things it’s heard before - by meaning.
+
+### Real-Time Session Cleanup – Redis Handles It
+Want to check what's still in memory?
+
+```redis:[run_confirmation=false] Check Sessions
+HGETALL session:42
+```
+
+Only the unexpired fields remain. Redis does the cleanup invisibly in the background.
+Your assistant has a clean, focused mind at all times.
+
+### Next Steps
+Now that your assistant has memory and meaning, you can:
+    - Tie session messages to store embeddings for per-session recall
+    - Use RAG (Retrieval-Augmented Generation) by combining Redis Vector Search with LLMs
+    - Add per-user memory: prefix session keys with a user ID (user:42:session:...)
+    - Introduce a fallback to persistent storage for long-term memory using Redis Flex