How to set up a vector index via sql #1419
Replies: 2 comments 1 reply
-
I am sorry for bothering you again, but work with the vector without java is it possible, impossible.... if it is impossible it is fine to me! At this point I would just like to know :) |
Beta Was this translation helpful? Give feedback.
-
Happy New Year, sorry for the delay. I responded on Discord, but I haven't seen a discussion here.
The documentation about vectors is still poor, there are so many things to cover and missing examples. We're going to improve this asap. For now, please check this source for more information: The SQL command creates the index automatically if not existent.
Yes, SQL command.
Open Cypher doesn't support vectors in the syntax. You can use it via SQL and Java API only.
These values can't be changed after creation (any vector dbms has the same limitation). If you need to change it, then drop and recreate the index. |
Beta Was this translation helpful? Give feedback.
-
Hi everyone, Happy New Year!
I have a query regarding vector information handling in ArcadeDB. While I've reviewed the documentation, I am afraid I need some additional details. I somewhat understand the Java code for creating an in-memory index for vectors, but as I'm not using Java, this example isn't directly applicable. The SQL example, as I understand it, assumes an existing index, as it is an import statement (import database https://dl.fbaipublicfiles.com/fasttext/vectors-crawl/cc.en.300.vec.gz with distanceFunction = 'cosine', m = 16, ef = 128, efConstruction = 128;).
Is there a method to create the index through the management interface or via SQL/cypher/gremlin..., ? That would enable me to do that via the Python wrapper library!
Specifically, I need to know how to:
-- persistentIndex = (HnswVectorIndex) database.getSchema().getIndexByName("Word[name,vector]");
-- List<SearchResult<Vertex, Float>> approximateResults = persistentIndex.findNeighbors(input, k);
Apologies if these questions seem basic or naive. If you have a comprehensive example encompassing the creation of a database, addition of an index, node insertion, and relationship assessment would be immensely helpful.
Best regards,
Daniel
Beta Was this translation helpful? Give feedback.
All reactions