How to order complex merges with a long topic list of lists #2334
Unanswered
Robs-Git-Hub
asked this question in
Q&A
Replies: 1 comment
It shouldn't. It just merges it all simultaneously.
I don't believe it does. It maps all of the topics to another in one go by simply merging everything from the topics together. |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
I would like to understand the best practice for merging complex topic lists of lists.
I have seen in past issues that it is best to provide a single list rather than split it up. In other words [-1, 3, 50] rather than
[-1, 3]
[-1, 50]
What I would like to know is (a) whether the order of items after the first in the list makes any difference, and (b) whether the order of lists within a list of lists makes any difference. If the merges are processed sequentially then having a lower id topic "merged out" from the set of topics earlier than a higher id topic could be problematic. If a list of pairs was unproblematic then that would be the easiest way to handle it, but the advice above suggests not to go that route and that inevitably means with a complex list of long lists that you'll end up with some lower id topics being merged before higher id ones.
Looking for any good practice advice for complex merge cases.
Thanks so much to the maintainers of BERTopic. It's excellent.
All reactions