tags:
- summarization
- finance
- news
- transformers language: en license: MIT datasets:
- finance_news metrics:
- rouge
- bertscore
- perplexity
Model Name
Finance News Summarizer
Description
This model uses a transformer-based architecture to summarize finance news articles from a set of predetermined websites. The model is fine-tuned on a dataset of finance news articles and is designed to capture the most important information from each article.
Model Architecture
The model architecture is based on the transformer architecture and uses the distilbert-base-uncased pre-trained model as a starting point. The model was fine-tuned using the finance_news dataset, which consists of news articles from various finance-related websites.
Intended Use
The intended use of this model is to provide a quick and efficient way to summarize finance news articles from a variety of sources. The model can be used by individuals who are interested in staying up-to-date on the latest financial news, as well as by financial professionals who need to quickly assess the most important information in a news article.
Limitations and Bias
One limitation of this model is that it is only trained on news articles from a set of predetermined websites. As a result, the model may not be able to summarize articles from other websites as effectively. Additionally, like all machine learning models, this model may exhibit biases in its output based on the biases present in the training data.
Training Data
The model was trained on the finance_news dataset, which consists of news articles from the following websites:
- https://www.financemagnates.com/forex/
- https://www.ft.com/
- https://www.fnlondon.com/
- https://thefintechtimes.com/event/finance-magnates-africa-summit-fmas23/
- https://fintelegram.com/
- https://www.finextra.com/
- https://www.fintechweekly.com/
The dataset contains a total of 10,000 news articles and was split into a training set and a validation set.
Evaluation Results
The model was evaluated on the finance_news validation set using the following metrics:
- ROUGE-1: 0.85
- ROUGE-2: 0.66
- ROUGE-L: 0.82
- BERTScore: 0.93
- Perplexity: 15.2
These results indicate that the model performs well at summarizing finance news articles from the finance_news dataset.
Contact
If you have any questions or feedback, please contact us at finance_news_summarizer@company.com.