POS Tagging (Part-Of-Speech)
Parse sentences according to grammar and determine the function of each term, e.g. John (noun) gave (verb) the (det) ball (noun) (01 Intro s.14). Part of advanced linguistic preprocessing, alongside word sense disambiguation (which sense a word has; normalizing synonyms like United States/USA/US and pronouns). Used by Concept - Lemmatization (s.19).
Related
Appears in
Use for feature selection (s.32) (01 Intro - Chunk 02 (s.31-48))
Filter tokens by POS tag: adjectives (JJ.) for sentiment analysis, nouns (N.) for text clustering (“red and blue cars are similar”). RapidMiner: PENN (English), STTS (German), regex conditions; Python NLTK: PENN. Slide shows a parse tree S -> NP VP for “that man read this book”. See Concept - Feature Selection.
From 02 Pre-processing, s.44-48
- Needed when the goal is specific (names of people, places, organisations; e.g. San Francisco). Tag sets range from 6-7 categories to nearly 100; minimum noun, verb, adjective, adverb, preposition, conjunction (s.44).
- Uses: feature reduction (only verb/adjective/adverb for sentiment); POS distribution for author, gender, genre (formal vs informal) classification (s.45).
- Concept - Penn Treebank: 36 tags, WSJ corpus; Brill tagger (public domain); Stanford Parser. Example “My/PRP$ dog/NN also/RB likes/VBZ eating/VBG sausage/NN ./.” (s.46-48).
- Related: Concept - Parsing, Concept - Word Sense Disambiguation. Source: 02 Pre-processing - Chunk 02 (s.31-60)