Resources Credentials to access slides:
username: nlp2019 password: 2019nlp
Notes taken by Luizo ( @LuigiBrosNin on Telegram)
Won’t be taking lessons this year, might do something if i take the exam B)
1.1 Intro
Summarization of the general paradigm of Computational Linguistics
Machine Learning dominates the NLP studies
If we embrace an empiricist view, then for analyzing a specific phenomenon X in a language variety Y (both manually or with computational tools):
- we have to collect a good set of textual data (a corpus) composed of texts from the language variety Y;
- we will collect all the occurrences of the phenomenon X from the corpus;
- we will analyse the occurrences of the phenomenon X to find regularities that will allow us to completely describe the phenomenon inside the corpus;
- we will generalize our conclusions, derived from corpus analysis, to the entire variety.
Zipf's law (1949) → a small portion of words in a language compose 40~% of each sentence
Linguistic levels:
- Phonetics/Phonology → how it sounds / shows for sign languages
- Morphology → the various forms words can have
- Infection (Derivations, compositions) (eg. cat/cats)
- Syntax → part of speeches, dependency, parsing, etc
- Semantics → Lexical semantics (synonymy, antonymy, etc)
- Discourse analysis / Pragmatics
- Anaphors: Mary is a woman. She loves Jhon.