As mentioned in the Docs, LexicalSyntacticFeaturizer creates lexical and syntactic features to support entity extraction. Also, CRFEntityExtractor has a parameter called “features” to extract features in the sliding window, and it only requires tokens as input (and dense_features optionally). Thus, this extractor doesn’t require a Featurizer. So when would the LexicalSyntacticFeaturizer be useful? I hope you might be able to give me some insights on this.