Financial News Sentences with Annotator Agreement Sentiment Labels

Encompassing sentences from English-language financial news categorized by sentiment. The categorization is based on the agreement rate of 5-8 annotators per sentence. The dataset size is estimated to be between 1,000 and 10,000 rows based on its '1K<n<10K' size category tag.

Use Cases

• Train a sentiment classification model on financial news sentences using the expert-generated sentiment labels.
• Analyze the relationship between annotator agreement rates and the sentiment categories assigned to sentences.
• Fine-tune a text classifier for the specific domain of financial news using the provided text and sentiment annotations.

Strengths

• Sentiment labels are expert-generated, not crowdsourced.
• Dataset size is categorized as between 1,000 and 10,000 rows.
• Each sentence's label incorporates agreement rates from 5-8 annotators.

Limitations

• The exact number of rows and columns is unknown.
• The dataset's temporal and geographic coverage are not specified.
• The license is listed as CC BY-NC-SA 3.0, which restricts commercial use.

Provenance

Source

rajistics on Hugging Face, originally from the Financial PhraseBank.

Collection Method

Sentences collected from financial news and categorized by sentiment with multiple annotators.

Freshness

Last updated on the platform in July 2022.

Geography

Region tag indicates 'us', but specific coverage is unknown.