zensols.nlparse.feature.lang
Feature utility functions. In this library, all references to panon stand for parsed annotation, which is returned from zensols.nlparse.parse/parse.
first-propbank-label
(first-propbank-label anon)Find the first propbank label across all .
first-sent-propbank-label
(first-sent-propbank-label anon)Find the first propbank label for a sentence.
mentions-feature-metas
(mentions-feature-metas entities)Return the feature metadata for mentions-features.
mentions-features
(mentions-features panon tokens entities)(mentions-features panon tokens entities mentions-keys)Return mentions features for panon with defaults of 0 for all named entities. The parameter mentions-keys are a list of mentions keys (defaults to :mentions) and can include custom mentions (i.e. token regular expression entities). Features are computed on the top level mentions comments returned include
- :mentions-count-E number of times entity E has occurred
- :mentions-ratio-E number of times entity E has occured / total tokens.
pos-tag-features
(pos-tag-features tokens)Return POS tag features computed all tokens, which include:
- :pos-first-tag The first POS tag.
- :pos-last-tag The last POS tag.
- :pos-tag-count-adjective: The number of adjectives found.
- :pos-tag-ratio-adjective: The ratio of adjectives to other POS tags.
- :pos-tag-count-adverb: The number of adverbs found.
- :pos-tag-ratio-adverb: The ratio of adverbs to other POS tags.
- :pos-tag-count-noun: The number of nouns found.
- :pos-tag-ratio-noun: The ratio of nouns to other POS tags.
- :pos-tag-count-verb: The number of verbs found.
- :pos-tag-ratio-verb: The ratio of verbs to other POS tags.
- :pos-tag-count-wh: The number of what/when/where type questions POS tags.
- :pos-tag-ratio-wh: The ratio of what/when/where type question POS tags to other POS tags.
- :pos-tag-count-adjective: The number of adjectives found.
- :pos-tag-ratio-adjective: The ratio of adjectives to other POS tags.
sentiment-features
(sentiment-features panon)Return sentiment features for parsed annotation panon, which include:
- :sentiment-utterance The utterance level sentimment score.
- :sentiment-sentence-sum The utterance sum across sentences.
- :sentiment-sentence-mean The sentiment mean across sentences.
- :sentiment-sentence-variance The sentiment variance across sentences.