[8.6] Adds machine learning release highlights. (#91903)

Author: szabosteve

docs/changelog/90450.yaml

@@ -0,0 +1,29 @@
+pr: 90450
+summary: Semantic search endpoint
+area: Machine Learning
+type: feature
+issues: []
+highlight:
+  title: Semantic search
+  body: |-
+    Semantic search uses an NLP model to generate a dense vector representation 
+    of the given query text. This vector, known as an embedding, is passed to 
+    kNN vector search to find related documents which are nearby in the vector 
+    space. Documents that are close together in vector space have a semantically 
+    similar meaning. This process makes it possible to find results that are not 
+    only lexically similar to the query but in the intent or the meaning of the 
+    text. For example the NLP model may understand that some words or phrases 
+    are essentially synonyms and unearth documents that would not be discovered 
+    by traditional lexical match. The potential of semantic search is boosted by 
+    combining kNN vector search with Elasticsearch queries in a hybrid retrieval 
+    strategy.
+
+    [discrete]
+    [[categorize-text-agg-ga]]
+    === The categorize text aggregation is generally available
+    
+    The {ref}/search-aggregations-bucket-categorize-text-aggregation.html[multi-bucket aggregation]
+    that can group semi-structured text into categories is generally available 
+    from 8.6.
+
+  notable: true