Conversion-based conversion-based labeling (Brill labeling)

What are the disadvantages and strengths of the Brill Tagger? Can you suggest some possible improvements for the tag?

+5
source share
2 answers

The biggest weakness of the Brill tags is the time required for the training phase (look at the timestamps for ACOPOST here or try to implement it with NLTK to get an idea). Remember that you should always consider the Brill tag as the last tag to be used in the sequence of tag systems (for simple marking, I usually use and train the Brill tag at the output of the HMM tag). Besides the fact that the training phase is even longer, using Brill tags in and of itself usually leads to a very large, usually overlapping, and sometimes “wrong” set of rules (i.e. Rules that trade in many “true” tags in many correct tags) .

Brill - , , , , , . , , , , ( , , , ). , .

, Brill , , ( , HMM). , Brill , , ​​ (.. , , Brill, , , ).

+7

Brill " : POS-" " ". , POS RDRPOSTagger Brill, , , . , RDRPOSTagger , Brill's. . .

+1

All Articles