Created
April 16, 2021 17:23
-
-
Save Lord-V15/c0b97d7546f271d6e5f86125323dd917 to your computer and use it in GitHub Desktop.
Settings and search calls in ElasticSearch for vector similarity search
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
# Idea : use this to index Quora's Question Pair dataset and explore the search results | |
doc = { | |
"settings": { | |
"index.knn": True | |
}, | |
"mappings": { | |
"properties": { | |
"title": { | |
"type" : "text" | |
}, | |
"title_vector": { | |
"type": "knn_vector", | |
"dimension": 128 | |
} | |
} | |
} | |
} | |
es.indices.create(index="questions",body=doc,ignore=400) # New index created using format specified above | |
def search_knn(title): | |
x = np.asarray(embed([title])).tolist()[0] # embed() creates embeddings of a sentence of 128 dimensions | |
script_query = { | |
"knn": { | |
"title_vector": { | |
"vector": x, | |
"k": 2 # K=2 so a maximum of 2-NN | |
} | |
} | |
} | |
response = es.search( | |
index="questions", | |
body={ | |
"size": 10, # Choose any size limit | |
"query": script_query, | |
"_source": {"includes": ["title"]} # Must have a title | |
} | |
) | |
return response |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment