Hello,
I've seen the webinar and towards the end, there was a mentioning that the baseline doesn't use any sort of machine learning. Is using Machine Learning a requirement for this challenge?
I know you stated that the baseline is basically creating the annotations by logic. I just wanted clarification to know if this has to have some sort of ML in it to distinguish the unstructured tables?
I'm not an expert with ML but I've thought of different approaches to possibly get the results without using ML. Also, for the tool we create can it import .tsv files from within the docker to help supplement producing a
desired output?
Could you please provide some clarification?
Respectfully,
Darin
Created by Darin decoderz Hello Darin,
There is no requirement for ML. You can use any approach that you wish.
Concerning importing .tsv files, the requirement for this challenge is that you annotate the unknowns with the metadata from the caDSR (supplemented if necessary with concepts from the NCI Thesaurus). Perhaps we'd need to know a bit more about how you plan to use those imported files, but in general we wouldn't disallow your use of other files for e.g. development & testing.
Regards,
Gilberto