Natural Language Processing

From Crowfly

Jump to: navigation, search

Back to AI Class Notes

Contents

Introduction

  • Language sets us apart from other animals and machines
  • Application - communicate with machines
  • Learning - much information is written down.

Language Models

Two types of models

  • Sequence of works
  • Trees

Bag of Words

Probabilistic Models

Markov Assumption means that affect on one variable on another is local.

Stationarity Assumption means that probability of each word is the same.

Language and Learning

Language in the world shows us many things

  • Trend of Ice Cream shows the Seasons of the Earth
  • Trend of Moon shows phaces of the moon.

Language Models Question ?

All but the last one is an example of problems that word models can be used to solve.

Unigram Model Samples

Bigram Model Samples

Trigram Model Samples

N Gram Model Samples

N Gram Model Question ?

Probability Question ?

Language Question ?

Letter Bigram Question ?

Trigram Model Question

Classification

Classification Question

Gzip

Segmentation



Back to AI Class Notes

Personal tools