Processed Parliamentary Speech Corpus from the United States

Over 1 million text records of parliamentary proceedings, processed and uploaded by HarrisDePerceptron to Hugging Face in February 2022. This dataset contains transcripts from U.S. government legislative sessions, formatted for machine learning applications. The data is stored in Parquet format and is compatible with libraries like Polars, Pandas, and Datasets.

Use Cases

• Train language models on political speech using the text transcripts.
• Analyze sentiment and topic trends within parliamentary debate text.
• Perform named entity recognition to identify mentioned politicians, organizations, and legislation in the speech corpus.
• Study temporal patterns in legislative language across different sessions of parliament.

Strengths

• Contains over 1 million text records, as indicated by the 'Size Categories1 Mn10 M' tag.
• Data is provided in the efficient Parquet file format for fast processing.
• Explicitly tagged for the U.S. region, providing clear geographic focus.

Limitations

• Specific column names and data schema are unknown, complicating direct analysis.
• The dataset was last updated in early 2022, which may not include recent parliamentary sessions.
• The original source and exact method of data collection are not documented.

Provenance

Source

U.S. parliamentary proceedings, inferred from 'Regionus' platform tag.

Collection Method

Processed from original transcripts; exact gathering method is unknown.

Freshness

Last updated on 2022-02-06; update frequency is unknown.

Geography

United States, as indicated by the 'Regionus' platform tag.