It decreases perplexity by providing clearer sentence boundaries

It increases the perplexity by introducing additional tokens in the vocabulary

We do not know how exactly, it only adjusts the total count of word tokens considered in the perplexity calculation

It has no impact on the perplexity calculation

A probabilistic model based on three-word combinations

A sequence of three words occurring together in a text

A syntactic parsing method focusing on triplets of words

A non-projective tree can be drawn with no crossing edges

Incompleteness of all treebanks automatically derived from phrase-structure treebanks

transition-based

graph-based

both

As an evaluation metric

As an embedding

As a target in the loss function

No

Yes

BPE tokenizer is not a machine learning model, so it cannot be trained

Can be used with both types of PoS models

Can be used only with CRF, because CRF is discriminative 

Can be used only with HMM, because HMM is a generative model

It depends

Yes

No

NLP requires large corpora

Natural languages have a regular nature

Permanent changes in natural languages