/toi

Primary LanguageJupyter Notebook

The Life and Times of The Times of India

Using the Proquest Times of India Corpus spanning 18XX--2008, we shed light on a slew of interesting questions.

Scripts

  • Parse ToI Parse XML to CSV. Here's the data dictionary

  • Basic Analyses

    1. number of articles per issue (by pub_date/year/month/weekday--weekend)
    2. number of words per article/title over time
    3. number (proportion) of articles by contributor w/ TNN (rest are presumably sourced via AP etc. but good to groupby)
    4. gender, religion etc. of contributors -- histogram of top 50 names, surnames using naampy and pranaam
    5. number of contributors per article
    6. number of editorial/news

  • NER

  • NER analyses

    1. Histogram of top 50 people covered.
    2. Histogram of top 50 places covered
    3. Gender, religion of people mentioned using naampy and pranaam

  • Other Ideas

    1. number of classified ads (on startpage == 1)
    2. number of ads (on startpage == 1)
    3. US vs. USSR/Russia etc.
    4. matrimonial ads: "caste no bar", 'fair complexion', etc.
    5. Need Annotation
      • news/not news
      • gov vs. not in ads
      • episodic'---x happened vs. 'thematic' --- more detailed/contextual piece
      • local/national/foreign news