dataset_info:
features:
- name: transcript
dtype: string
- name: sentiment
dtype: string
splits:
- name: test
num_bytes: 182442
num_examples: 700
download_size: 98661
dataset_size: 182442
configs:
- config_name: default
data_files:
- split: test
path: data/test-*
license: mit
task_categories:
- text-generation
language:
- en
tags:
- finance
- financial sentiment
size_categories:
- n<1K
Aiera Financial Sentiment Analysis Dataset
Description
This dataset focuses on the sentiment analysis of earnings call transcript segments. It provides pre-segmented extracts from earnings calls, transcribed by Aiera, paired with sentiment labels. Each segment in the transcript column is annotated with a sentiment label (sentiment), which can be "positive", "negative", or "neutral". This dataset is intended for training and evaluating models on their ability to discern the underlying sentiment in financial communications.
Dataset Structure
Columns
transcript: A segment of the earnings call transcript.sentiment: The sentiment label for the transcript segment, with possible values being "positive", "negative", or "neutral".
Data Format
The dataset is structured in a tabular format, with each row representing a unique segment of an earnings call transcript alongside its corresponding sentiment label.
Use Cases
This dataset is particularly suited for applications such as:
- Training machine learning models to perform sentiment analysis specifically in financial contexts.
- Developing algorithms to assist financial analysts and investors by providing quick sentiment assessments of earnings calls.
- Enhancing natural language processing systems used in finance for better understanding of market mood and company performance.
Accessing the Dataset
To access this dataset, you can load it using the HuggingFace Datasets library with the following Python code:
from datasets import load_dataset
dataset = load_dataset("Aiera/aiera-transcript-sentiment")
A guide for evaluating using EleutherAI's lm-evaluation-harness is available on github.