Merge branch 'redis:main' into main

README.md

@@ -1,10 +1,14 @@
-This module implements vector sets for Redis, a new Redis data type similar
-to sorted sets but having a vector instead of a score. It is possible to
-add items and then get them back by similiarity to either a user-provided
-vector or a vector of an element already inserted.
+This module implements Vector Sets for Redis, a new Redis data type similar
+to Sorted Sets but having string elements associated to a vector instead of
+a score. The fundamental goal of Vector Sets is to make possible adding items,
+and later get a subset of the added items that are the most similar to a
+specified vector (often a learned embedding) of the most similar to the vector
+of an element that is already part of the Vector Set.
 
 ## Installation
 
+Buil with:
+
     make
 
 Then load the module with the following command line, or by inserting the needed directives in the `redis.conf` file.
@@ -19,7 +23,7 @@ The execute the tests with:
 
     ./test.py
 
-## Commands
+## Reference of available commands
 
 **VADD: add items into a vector set**
 

hnsw.c

@@ -648,7 +648,11 @@ pqueue *search_layer(HNSW *index, hnswNode *query, hnswNode *entry_point,
 
         /* Stop if we can't get better results. Note that this can
          * be true only if we already collected 'ef' elements in
-         * the priority queue. */
+         * the priority queue. This is why: if we have less than EF
+         * elements, later in the for loop that checks the neighbors we
+         * add new elements BOTH in the results and candidates pqueue: this
+         * means that before accumulating EF elements, the worst candidate
+         * can be as bad as the worst result, but not worse. */
         float furthest = pq_max_distance(results);
         if (cur_dist > furthest) break;