search

tarahmarie / dh-trace

MIT License
0 stars 0 forks source link

Next issue on tuning data sources #11

Closed tarahmarie closed 2 weeks ago

tarahmarie commented 1 year ago

1) Tune the variables for most effective weights for highest accuracy - F1 score is how to make that work (graph to visually confirm) 2) compare ngrams to SVMs to see if SVMs gives a better numeric result 3) Regenerate alignments after considering updating 4) consider getting the base text files to be the same as those used in alignments

jdmartin commented 1 year ago

On point 4: Not 'consider,' so much as 'absolutely do this for the sake of sanity'. ;)

tarahmarie commented 2 months ago

Are all 4 of these issues outstanding? The major one is the SVMs issue next.

tarahmarie commented 2 weeks ago

This issue is outdated and 90% complete; closing in favor of further specificity.