Word cloud

Word cloud is a visual representation of n-grams that give greater importance to words that appear more frequently in a source text. The bigger and bolder the n-gram appears, the more frequently it appears in the text.

Graph display unigrams (single words), bigrams, and trigrams for the source text.

Coding example:

Use case: Essential topics in newspaper headlines

Data: Million News Headlines dataset, source: Australian Broadcasting Corporation, data licence: CC0 1.0: Public Domain.

Coding:

1import pandas as pd
2from arabica import cappuccino

1 data = pd.read_csv('abcnews_data.csv', encoding='utf8')

The data looks liks this:

headline

date

aba decides against community broadcasting licence

2003-2-19

act fire witnesses must be aware of defamation

2003-2-19

It procceeds in this way:

  • additional stop words cleaning, if skip is not None

  • lowercasing: reviews are made lowercase so that capital letters don’t affect n-gram calculations (e.g., “Tree” is not treated differently from “tree”), if lower_case = True

  • punctuation cleaning - performs automatically

  • stop words removal, if stopwords is not None

  • digits removal, , if numbers = True

  • n-gram frequencies for each headline are calculated, summed, and displayed in a word cloud.

 1cappuccino(text = data['headline'],
 2           time = data['date'],
 3           date_format = 'us',                 # Uses US-style date format to parse dates
 4           plot = 'wordcloud',
 5           ngram = 2,                          # N-gram size, 1 = unigram, 2 = bigram, 3 = trigram
 6           time_freq = 'ungroup',              # No period aggregation
 7           max_words = 150,                    # Displays 150 most frequent bigrams
 8           stopwords = ['english'],            # Remove English stopwords
 9           skip = ['covid','great day today'], # Remove additional stop words
10           numbers = True,                     # Remove numbers
11           lower_case = True)                  # Lowercase text

Here is the output:

alternate text

Download the jupyter notebook with the code and the data here.