benwtrent commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2108168850
> Not sure I understand option 3. Are you thinking that graph has different
types of edges b/w documents based on diff. similarity functions? So if you
were using max similarity y
vigyasharma commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2108154488
@benwtrent I like the idea of having documents be vertices in the graph,
with an API that let's you iterate/access the different vectors per doc. It
would have an indexing time
benwtrent commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2107660631
@vigyasharma @krickert There are a couple of ways to implement this natively
in Lucene.
1. Have each individual vector be a connection in the graph with some
resolution ba
vigyasharma commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2107568883
> In another scenario, the results would just return the top doc and not
repeat it.
I believe this is what the parent-block join implementation for vector
values does cur
krickert commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2105745361
I was thinking about this and thought this would be cool with a few
different use cases for a multi-valued vector:
1. The multi-values are treated the same as the single valu
benwtrent commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2045722371
> if the aggregation is max, would we need to compute distance between n x
n vectors and then take the max?
Correct, I would even want flexibility between what was used to
vigyasharma commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2045706297
@benwtrent : ++, I've been thinking on similar lines in the context of
e-commerce type applications where different vectors represent different
aspects of a document. The scorer
benwtrent commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2045604262
I do think things like `ColBERT` would benefit from having multiple vectors
for a single document field.
One crazy idea I had (others have probably already thought of this,
alessandrobenedetti commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2044685997
Hi, I gave a talk about this at Berlin Buzzwords where I touched on the
motivations:
https://www.youtube.com/watch?v=KhL0NrGj0uE
In short:
- multi-valued vectors
vigyasharma commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2044076948
What are some use-cases for multi-valued vectors that are not easily
supported using parent-child block joins?
I'd like to contribute here, trying to understand our main
alessandrobenedetti commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-1831534483
Hi @david-sitsky, the multi-valued vectors in Lucene's contribution is now
paused for lack of fundings.
I'll resume it from my side if and when I get some sponsors :)
david-sitsky commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-1831197772
> The key issue is document collection. Right now, the `topK` is limited to
only `topK` children documents. Really, what you want is the `topK` parent
documents based on childr
12 matches
Mail list logo
