[‘TextBlob‘, ‘is‘, ‘a‘, ‘Python‘, ‘library‘, ‘for‘, ‘processing‘, ‘textual‘, …]

Introduction

Natural Language Processing, or NLP for short, is a fascinating field of artificial intelligence focused on enabling computers to understand, interpret, and generate human language. NLP powers many applications we use every day, from virtual assistants like Siri and Alexa that can engage in human-like dialog, to machine translation between languages, to analyzing the sentiment of social media posts and customer reviews.

While natural language processing is a complex field of study, thanks to open source libraries, it‘s easier than ever for beginners to get started with NLP. In the Python world, one of the most beginner-friendly NLP libraries is TextBlob. Built on top of the popular NLTK library, TextBlob provides an intuitive interface for performing common NLP tasks.

In this guide, we‘ll walk through how to use TextBlob for natural language processing, from basic setup to key concepts and techniques. Whether you‘re an aspiring data scientist looking to analyze unstructured text data, or a developer interested in building chatbots and other language-aware applications, TextBlob is a great starting point. Let‘s dive in!

Setting Up TextBlob

Before we can start processing text with TextBlob, we need to install it. Luckily, TextBlob is available on PyPI and can be easily installed with pip:

pip install -U textblob

TextBlob also requires downloading some additional resources, such as corpora and trained models. We can download these with the following command:

python -m textblob.download_corpora

And that‘s it – we‘re ready to start using TextBlob! We can verify the installation by opening a Python REPL and importing the library:

from textblob import TextBlob

If no errors occur, we‘re good to go.

Key NLP Tasks with TextBlob

Now that we have TextBlob installed, let‘s explore some of the most important NLP tasks we can accomplish with this versatile library.

Tokenization

Tokenization is the process of splitting text into smaller units, such as individual words or sentences. This is a foundational step in almost all NLP pipelines. With TextBlob, tokenizing text is very straightforward:

text = "TextBlob is a Python library for processing textual data. It provides a simple API for diving into common natural language processing tasks."
blob = TextBlob(text)
print(blob.words) 

print(blob.sentences)

As you can see, accessing the .words and .sentences properties of a TextBlob object returns lists of tokens. TextBlob handles tokenization intelligently, including not splitting contractions like "can‘t".

Part-of-Speech Tagging

Part-of-speech tagging, or POS tagging, is the process of labeling each token in the text with its part of speech (noun, verb, adjective, etc). Knowing the parts of speech can be useful for tasks like named entity recognition and analyzing sentence structure.

To get the POS tags for a blob of text, use the .tags property:

blob = TextBlob("TextBlob is easy to use.")
print(blob.tags)  
# [(‘TextBlob‘, ‘NNP‘), (‘is‘, ‘VBZ‘), (‘easy‘, ‘JJ‘), (‘to‘, ‘TO‘), (‘use‘, ‘VB‘)]

The tags are represented as two-letter codes. For example, NNP stands for singular proper noun, JJ for adjective, VB for verb, etc. You can find a full list of POS tags and their meanings here.

Noun Phrase Extraction

Another common NLP task is extracting noun phrases – phrases that contain a noun and any modifiers. This can be useful for quickly summarizing the key topics in a piece of text. With TextBlob, extracting noun phrases is just one line of code:

blob = TextBlob("John is learning natural language processing with TextBlob in Python.")
print(blob.noun_phrases)
# [‘john‘, ‘natural language processing‘, ‘textblob‘, ‘python‘]  

As you can see, TextBlob‘s noun phrase extraction returns a list of the main noun phrases in the text, ignoring punctuation, case, etc.

Word Inflection and Lemmatization

In natural language, words often take on different forms through inflection – for example, pluralization of nouns or conjugation of verbs. Lemmatization is the process of reducing a word to its base or dictionary form (its lemma). TextBlob makes it easy to work with these word forms.

from textblob import Word
word1 = Word("octopi") 
word2 = Word("went")
print(word1.singularize())  # octopus
print(word2.lemmatize("v")) # go

In the first example, we singularize the plural noun "octopi" to get its singular form "octopus". In the second, we lemmatize the past-tense verb "went" to get its infinitive form "go", by specifying the "v" (verb) part of speech.

N-grams

An n-gram is a contiguous sequence of n items from a text. N-grams are used in NLP for tasks like language modeling and text similarity analysis. Getting n-grams with TextBlob is trivial:

  
blob = TextBlob("NLP is fun and useful!")  
print(list(blob.ngrams(n=2)))
# [WordList([‘NLP‘, ‘is‘]), WordList([‘is‘, ‘fun‘]), WordList([‘fun‘, ‘and‘]), WordList([‘and‘, ‘useful‘])]

Here we extract bi-grams (2-grams) from the text as a list of WordLists (a TextBlob container class). We can also get tri-grams, 4-grams and so on by changing the n parameter.

Sentiment Analysis

One of the most powerful features of TextBlob is its built-in sentiment analysis model. With just a few lines of code, we can get the sentiment polarity and subjectivity of a piece of text:

positive_blob = TextBlob("This restaurant is amazing!")
print(positive_blob.sentiment)
# Sentiment(polarity=0.6000000000000001, subjectivity=0.9)
negative_blob = TextBlob("This food is disgusting.")

print(negative_blob.sentiment)

Polarity is a float between -1 and 1, representing how positive or negative the sentiment is. Subjectivity, a float between 0 and 1, shows how subjective (as opposed to factual) the text is. TextBlob makes it easy to gauge the sentiment of customer reviews, social media posts, and more with this simple API.

Additional TextBlob Features

Beyond the core NLP tasks covered above, TextBlob provides several other useful features for working with text.

Spelling Correction

TextBlob can automatically correct spelling errors using a simple function:

blob = TextBlob("I havv goood speling!")
print(blob.correct()) 
# TextBlob("I have good spelling!")

The autocorrect functionality is powered by the Peter Norvig spelling corrector, which uses word frequency statistics to guess the most likely correct spelling.

Language Detection and Translation

Another nifty feature of TextBlob is its language detection and machine translation capabilities, powered under the hood by Google Translate:

 
blob1 = TextBlob("Bonjour, comment allez-vous?")
print(blob1.detect_language()) # fr
blob2 = blob1.translate(to="en") 
print(blob2) # TextBlob("Hello how are you?")

Here we have a French phrase that TextBlob correctly identifies as French, then translates into English. This functionality supports dozens of languages.

Text Summarization

While TextBlob doesn‘t have a built-in text summarization function, we can combine some of its other features to create a rudimentary extractive summary:

from textblob import TextBlob
import nltk
text = "The quick brown fox jumps over the lazy dog. The quick brown fox is quick. The lazy dog is lazy."
blob = TextBlob(text)

topics = blob.noun_phrases

content_words = [w for w in blob.words if w.lower() not in nltk.corpus.stopwords.words("english")]
freq_dist = nltk.FreqDist(content_words)
key_words = [word[0] for word in freq_dist.most_common(5)]
print(f"Key Topics: {‘, ‘.join(topics)}")  

print(f"Key Words: {‘, ‘.join(key_words)}")

Here we use TextBlob‘s noun phrase extraction to get the key topics, and its part-of-speech tagging combined with NLTK‘s stopword list and frequency distribution to get the most frequent substantive words. This gives us a rough sense of what the text is about.

Of course, this is a very basic approach to summarization. For more advanced extractive and abstractive summarization techniques, you‘ll likely want to explore dedicated libraries like Gensim or deep learning models like BART and T5. But this example shows how we can combine TextBlob‘s features in creative ways.

TextBlob Pros and Cons

As we‘ve seen, TextBlob is a powerful and easy-to-use library for NLP. However, it‘s important to understand its strengths and limitations compared to other popular libraries like NLTK and spaCy.

Pros of TextBlob

• Beginner-friendly, intuitive API
• Built-in functionality for common NLP tasks
• Includes useful features like sentiment analysis, spelling correction, and language detection/translation
• Faster than NLTK for many tasks

Cons of TextBlob

• Less flexible and customizable than NLTK
• Lacks advanced features of spaCy like named entity recognition and dependency parsing
• Slower than spaCy for large scale text processing
• Sentiment model is less accurate than more advanced ML approaches

Overall, TextBlob shines as an approachable option for beginners and for rapidly prototyping NLP applications. As you scale to larger, more complex tasks, you may find yourself reaching for the greater customization of NLTK or the speed and advanced features of spaCy. The best approach is to understand the tradeoffs and choose the right tool for each job.

Conclusion

We‘ve covered a lot of ground in this guide to natural language processing with TextBlob! We installed the library, learned key NLP concepts and techniques including tokenization, part-of-speech tagging, lemmatization, n-grams, and sentiment analysis, and explored bonus features like spelling correction, language translation, and basic text summarization.

If you made it this far, congratulations – you now have a solid foundation for working with text data in Python. Of course, we‘ve only scratched the surface of the vast field of NLP. To further your learning, I recommend diving deeper into the TextBlob documentation, exploring NLTK and spaCy, and studying fundamental NLP concepts like bag-of-words models, TF-IDF, topic modeling, and embeddings.

At the end of the day, the best way to get better at NLP is to practice. Try extracting insights from your own text datasets, building a chatbot, or even training your own custom text classifier. The possibilities are endless!

I hope this guide has been helpful for you. Feel free to reach out with any questions, and happy NLP-ing!