README.md

metadata

language:
  - ar
license: apache-2.0
task_categories:
  - multiple-choice
  - text-classification
pretty_name: Sentiment Analysis MCQ Evaluation Dataset
tags:
  - sentiment-analysis
  - mcq
  - financial
  - arabic
  - evaluation
  - benchmark
configs:
  - config_name: default
    data_files:
      - split: test
        path: data/test-*
dataset_info:
  features:
    - name: original_split
      dtype: string
    - name: choices
      sequence: string
    - name: text
      dtype: string
    - name: answer
      dtype: string
    - name: original_sentiment
      dtype: string
    - name: query
      dtype: string
    - name: id
      dtype: string
    - name: gold
      dtype: int64
    - name: category
      dtype: string
  splits:
    - name: test
      num_bytes: 495875
      num_examples: 80
  download_size: 218370
  dataset_size: 495875

Sentiment Analysis MCQ Evaluation Dataset

Validation and test splits for financial sentiment analysis in MCQ format.

Dataset Structure

• Format: Multiple choice questions
• Language: Arabic
• Domain: Financial reports
• Task: Sentiment classification
• Validation: 20 examples
• Test: 20 examples

Fields

• id: Unique identifier
• query: Full MCQ prompt
• answer: Correct answer letter
• text: Question text
• choices: Answer options [a, b, c]
• gold: Correct answer index
• category: Report category
• original_sentiment: Ground truth sentiment

Answer Mapping

• a) positive - gold: 0
• b) negative - gold: 1
• c) neutral - gold: 2

Usage

from datasets import load_dataset

dataset = load_dataset("SahmBenchmark/Sentiment_Analysis_MCQ_eval")
test_data = dataset['test']

for example in test_data:
    print(f"Question: {example['text']}")
    print(f"Choices: {example['choices']}")
    print(f"Correct: {example['answer']} (index: {example['gold']})")