Analysis of Wikipedia Article Length Across Languages

Objective

This project explores whether English Wikipedia articles tend to be longer and richer in content compared to their French counterparts.

Methodology

To investigate, I analyzed the length of Wikipedia articles in various languages. I used the Wikipedia API to collect data. Due to API limitations (50 random articles content at a time), I first calculated the average bytes per character using this API, then used another more permissive API to fetch article lengths (500 articles at a time). This approach enabled me to determine the number of characters for a large set of articles.

Key Findings

On average, English Wikipedia articles contain 19% more information than French articles.

Getting the Dataset

1. Download 100 000 articles and their lengths without content for each studied language:

   python3 scripts/fetch_random_articles.py

2. Download 500 articles with content and their lengths for each studied language:

   python3 scripts/fetch_random_articles_content.py

Processing the Data

Pre-processed data is available in the data/processed folder.

1. Process the dataset to get the average byte sizes of articles for each language:

   python3 scripts/process_articles_byte_size.py

2. Process the dataset to get the average bytes per character for each language:

   python3 scripts/process_articles_bytes_per_char.py

These commands will generate two files:

• data/processed/articles_byte_size.csv: Contains statistics (average, standard deviation, mean, 1st, and 3rd quartiles) of article byte sizes for each language.
• data/processed/articles_bytes_per_char.csv: Contains statistics of bytes per character for each language.

Analysis

For detailed results, see notebooks/analyis.ipynb.