This document provides a summary of machine learning techniques for natural language processing (NLP). It introduces topics like machine learning pipelines, tokenization, bag-of-words models, topic modeling and libraries for implementing NLP models. Topic modeling is defined as a method to analyze large volumes of text by clustering words that frequently occur together into topics. The document outlines common NLP preprocessing steps and machine learning algorithms for tasks like classification and clustering.

1. General guide in NLP
微光國際資訊有限公司 2018/05/16

2. About this chapter
• Introduction - Machine Learning in NLP
• Pipelines
• Tokenization
• Bag of Words and TF - IDF
• Topic Model
• Library - implement
• Visualization

3. Introduction - Machine Learning in NLP

4. Introduction - Machine Learning in NLP
Machine Learning
Supervised UnSupervised
•K-mean
•Clustering
•Topic Model
•K-NN
•Logistic regression
•Decision tree
•Classification

5. Topic Model App

6. –mallet CS In UMASS
Let's define Topic model
“Topic models provide a simple way to
analyze large volumes of text. A
"topic" consists of a of words that
occur together.”
unlabeled
cluster
frequently

7. Pipelines
Document
word
preprocess vector Model Visualization
• tokenization
• stopword
• lemm
• argument
• Bow
• Dictionary
•LDA
•LSA
•PLSA

8. TokenizationThe quick brown fox jumps over the lazy dog.The quick brown fox jumps over the lazy dog.jump

9. Bag of Words and TF - IDFquick brown fox jump over lazy dog
:1
:2
:3
:4
:5
:6
:7
Document Document Document
0
1
4
4
3
0
7
12
0
1
2
9
3
1
10
5
20
4
0
29
3
維度

10. Bag of Words and TF - IDF
Term Frequency Inverse Document FrequencyX
文字出現頻率 權重值

11. Topic ModelIn Math

12. 主題權重值參數
文字於主題權重值
參數
文件
主題數
文字
文字總量

14. Library - implement

15. Visualization