neuml / cord19q Goto Github PK

Hi David,
Thank you for helping us run paperai v1.2.1. successfully. Unfortunately, this version does not include embeddings.py, scoring.py, pipeline.py etc., which significantly improve the results’ quality.
So I tried to install cord19q v2.3.0 because we need to reproduce the high quality results you obtained with previous cord19q versions. However, I’m getting the following error, which might be due to the recent changes you made to neuml/magnitude

Do you think I should try to install cord19q v2.3.0 with a previous version of magnitude (i.e. neuml/magnitude v0.1.143)?
or would you be able to do something on your side that would allow me to install and run cord19q v2.3.0?

It would help the curators to have separate columns for title and url. Author no longer needs to be displayed.

It would be great for the cord19q package on GitHub if you clear the output before printing a new line for the etl, vector, and index functions. The output would be much cleaner without the thousands of lines scrolling on display. Thanks.

in etl.execute.py
for _, text in sections:
# Look for at least one keyword match
if re.findall(regex, text.lower()):
tags = "COVID-19"

if we find a "COVID-19" tag, we should return? No need to look through the other sentences?

Also, in execute.files, good to add a check for key not present in the dictionary:
if row.get('has_pmc_xml_parse'):
if row["has_pmc_xml_parse"].lower() == "true":

Thanks so much for making this code public. Learning so much from going through this.