Analyzing Traditional and Neural Approaches to Multilingual Readability Assessment
This work tests whether transformers internalize the same features as traditional models across Arabic, English, French, Hindi, and Russian using the ReadMe++ dataset, and identifies the features driving traditional classifiers.
J. Wong, Chris Tanner
· 0 citations