How This Workflow Works
This workflow analyzes financial news articles by prompting a large language model to classify each article's sentiment. It then evaluates the model's predictions against known sentiment labels, generates a PDF report summarizing the results, and distributes the report by email.
Key Features:
- Automate sentiment classification of financial news and company press releases using OpenAI’s LLMs
- Evaluate model performance against ground truth sentiment labels
- Generate and distribute a static PDF report summarizing sentiment analysis results
- Streamline reporting by sending results directly via email
Step-by-step:
1. Classify Sentiment Using a Language Model:
The workflow prompts OpenAI’s LLM to analyze each financial news article and assign a sentiment label—positive, negative, or neutral. This step leverages a carefully constructed prompt and authenticated access to the language model provider.
2. Evaluate Sentiment Predictions:
After classification, the workflow compares the model's sentiment predictions with the actual sentiment labels present in the dataset. This evaluation provides a clear measure of the model's accuracy and reliability.
3. Generate and Distribute a Sentiment Report:
The results of the sentiment analysis and evaluation are compiled into a static PDF report. The workflow then sends this report by email, ensuring that stakeholders receive timely updates without manual intervention.