Elastiflix-Movies Java

Vector component · k-NN

256 dimensions. It’s one of the three vector representations of the plot (plot). More dimensions usually capture more nuances of meaning, at the cost of more memory and more expensive comparisons.

Maximum number of results the vector search returns, after merging the candidates from all shards; they’re split into pages of 25, 50 or 100. It must be less than or equal to num_candidates. In hybrid mode it also sets how many results from each list go into the fusion and how many combined results you can page through.

Candidates evaluated on each shard before keeping the best k. Raising it tends to improve accuracy, at the cost of longer response times. Maximum 10,000.

k-NN (k nearest neighbors): your query is turned into a vector, and the search looks for the movies whose plot has the closest vector. Elasticsearch does this approximately over an HNSW graph, without comparing against the whole catalog, which makes it fast but not always exact.