Dear community,
I would like to calculate cosine similarity values using gds.similarity.cosine
and return top k elements.
Is it possible to use the using apoc.periodic.iterate
to parallelize this process and make calculation faster?
MATCH (h0:Hash)-[:Table0]->(vector0:Vector)
WHERE id(h0) = 4
with vector0 as vector, gds.similarity.cosine(vector0.vector, [0.348, 0.1900]) as cosineSimilarity
where cosineSimilarity>0.5
RETURN vector, cosineSimilarity
LIMIT 10
Is there maybe another idea how can I get KNN faster (new data are constantly added to Hash node)?