4 ms·
Cool idea! Shouldn't this work also by concatenating the single document (you want to classify) with the compressed version of the conc. class corpus (saving co
by sean_pedersen 5y ago
Cool idea! Shouldn't this work also by concatenating the single document (you want to classify) with the compressed version of the conc. class corpus (saving compute time)?
- Lemaxoxo 5y agoI think that is what is being suggested in the other comment. One would have to try! My instincts tell me the results would not be identical.
- woliveirajr 5y agoThere are some "old" research about it... Normalized compression distance, for example, is a good start about this approach. I've used this to classify documents in categories or even to find out the author of a document (0) (0)https://pubmed.ncbi.nlm.nih.gov/23597746/ https://pubmed.ncbi.nlm.nih.gov/23597746/