机器学习读书笔记（八）：Sentiment Analysis (Natural Language Processing i.e. NLP)

学习目标：

• Cleaning and preparing text data
• Building feature vectors from text document
• Training a machine learning model to classify positive and negative reviews
• Working with large text datasets using $out$-$of$-$core$ learning

1.Obtaining IMDb movie review dataset

Introductions:

1. Sentiment analysis is sometimes also called opinion mining
2. IMDb(Internet Movie Database) collected by Maas et al.

Actually, the author Sebastian Raschka uses a quite fancy way to downlowd the files and comments conveniently through os method, which may be not so convenient to program… Therefore we just skip this process and get our movie_data.csv as follow:

2.Bag-of-words model

Bag-of-words model allows us to represent text as numerical feature vectors, which can be summaried as follows:

1. Create a vocabulary of unique tokens
2. Construct a feature vector from the counts of words

Note that the feature vectors are actually sparse as we may have mentioned bofore,which means they consist of mostly zeros.

2.1Transforming words into feature vectors

#调整用n-gram
import numpy as np
from sklearn.feature_extraction.text import CountVectorizer

count = CountVectorizer()
docs = np.array([
        'The sun is shining',
        'The weather is sweet',
        'The sun is shining, the weather is sweet, and one and one is two'])
bag = count.fit_transform(docs)
print(count.vocabulary_)
print(bag.toarray())
--------------------------------------------------
{
   'the': 6, 'sun': 4, 'is': 1, 'shining': 3, 'weather': 8, 'sweet': 5, 'and': 0, 'one': 2, 'two': 7}
[[0 1 0 1 1 0 1 0 0]
 [0 1 0 0 0 1 1 0 1]
 [2 3 2 1 1 1 2 1 1]]

3.Accessing word relevancy via term frequency-inverse document frequency

tf-idf, short for term frequency-inverse document frequency, is defined as the product as follows:
$$tfidf(t,d)=tf(t,d)\times idf(t,d)$$
where $tf(t,d)$ means the number of times a term $t$ occurs in a document $d$.
And here, $idf(t,d)$ can be calculated as :
$$idf(t,d)=log\frac{n_d}{1+df(d,t)}$$
where $n_d$is the total number of documents, and d f (