2.0 Added missing speeches

• added missing speeches from candidates who where president in the election period

1.5 Filtered Speeches

• deleted all speeches outside of an election period and all speeches within an election period for speakers that were not candidates in that period
• speeches were reduced from 964 to 754

1.4 Small Update

• new Metadata columns tokens and sentences
• new corpus statistic plot
• sentiment analysis plot per election period with t test

V.1.3 New Corpus Structure & Plots

Changelog

• changed corpus structure from xml to csv files per speech
• csv file contains the following values: IDs, TOKEN, LEMMA, POS, MORPH, HEAD, DEPREL, NAMED_ENTITY, SENTIMENT_SENTENCE
• csv & text files are linked via metadata excel file
• added party membership and election period to metadata excel
• metadata plots for speeches per candidate, period and state
• sentiment analysis per speaker and year

Examples

Added XML

• added xml containing the corpus in TEI Lite standard

1.1 Removed Interviews

• removed 109 speeches with[Ii]nterview substring inside title
• 964 speeches remained

1.0 Raw result of web scraping

Result from scraping presidency.ucsb.edu on 10.11.2024. It contains 1073 speeches.

• The search url was https://www.presidency.ucsb.edu/advanced-search?field-keywords=&field-keywords2=&field-keywords3=&from%5Bdate%5D=01-01-2008&to%5Bdate%5D=11-08-2024&person2=&category2%5B0%5D=63&items_per_page=100&f%5B0%5D=field_docs_attributes%3A205
• The speakers included where ['John McCain', 'Barack Obama', 'Mitt Romney', 'Hillary Clinton', 'Donald J. Trump', 'Joseph R. Biden, Jr.', 'Kamala Harris'].
• It excluded results with 'Press Release' substring within the speech's title.

The zip file includes

• a folder containing all speeches as plain txt files
• a metadata table with one row per speech as csv & excel file
• a txt file with all urls which were scraped
• a quick png visualization of the metadata