RRF limited to 1000 docs?

[lintool]

RRF allows user to specify max_docs:

trectools/trectools/fusion.py

Line 106 in 9e76486

def reciprocal_rank_fusion(trec_runs, k=60, max_docs=1000, output=sys.stdout):

However, it seems only up to 1000 hits:

trectools/trectools/fusion.py

Line 123 in 9e76486

docs_for_run = r.get_top_documents(topic, n=1000)

Shouldn't max_docs be passed into n?

[x65han]

I got this

[joaopalotti]

Thanks, Jimmy for reporting it!
Johnson Han, thanks for taking care of it. I really appreciate it!
Looking forward to receiving your pull request.

[joaopalotti]

I had a look at this issue and we do not necessarily want to have:
docs_for_run = r.get_top_documents(topic, n=max_docs)

The parameter max_docs is only used for the final list of documents after the merging. Another parameter could be created to filter the top X documents of each run, if this is a use case.

Closing this issue for now.