Why is Natural Language Processing important?

Class 1 — July 09, 2020

NLP is Everywhere

Automatic Summarization: generating short summaries of articles
- Not quite there yet, but it’s getting better
Language Modeling: predicting what follows a given sequence of words/characters
- e.g. next word prediction on smartphone keyboards
Machine Translation: translating content between languages
- e.g. Google Translate
Question Answering: answering questions about a body of text
- Also not very good yet, but getting better
Speech Recognition: transcribing freeform speech
- e.g. Siri, Google Assistant, Alexa, rev.ai
Speech Synthesis: generating synthetic voices that read text aloud
- e.g. Siri, Google Assistant, Alexa, Polly
Text Classification: sorting bodies of text (Tweets, emails, etc.) into categories
- e.g. sentiment analysis, spam detection
Text Generation: generating text, usually in response to a prompt
- e.g. chatbots, GPT-2, GPT-3
and many, many more

NLP (Natural Language Processing): study of how computers can process and manipulate human languages
dataset: a large body of data that can be used to train a model
corpus: usually a very large, unstructured textual dataset
training: providing a machine learning model with data to learn from
testing: evaluating how well the model learned from your training data

Probability notation
- P(A) = probability of A occurring
- P(B | A) = probability of B occurring, given that A has already happened
(Basic) Linear algebra
- What is a vector? → list of numbers with spatial meaning
- How can you measure similarity between two vectors? → Euclidean distance, cosine similarity