DecisionQE

DecisionQE is a small English multiple-choice question-answering dataset for evaluating decision-making, persuasion, communication, and influence-related knowledge.

The dataset contains 70 questions across six practical domains. Each item includes a category, a question, multiple-choice options, and the correct answer key.

Dataset Files

• DecisionQE_dataset.json: merged dataset with metadata and all questions.

Dataset Structure

Top-level structure:

{
  "name": "DecisionQE",
  "title": "DecisionQE Dataset",
  "description": "Merged DecisionQE question-answer dataset from six testing domains.",
  "total_questions": 70,
  "source_summaries": [],
  "questions": []
}

Each question has the following fields:

Field	Type	Description
dataset	string	Dataset name, always DecisionQE.
id	integer	Sequential item id in the merged dataset.
category	string	Domain/category of the question.
original_id	integer or null	Original question id before merging.
question	string	Multiple-choice question text.
options	object	Answer options keyed by letters such as a, b, c, d.
answer	string	Correct option key.

Example:

{
  "dataset": "DecisionQE",
  "id": 1,
  "category": "interpersonal_communication",
  "original_id": 16,
  "question": "Research shows that to effectively build rapport in a conversation, you should:",
  "options": {
    "a": "Mirror the other person's body language and speech patterns",
    "b": "Maintain a completely neutral expression",
    "c": "Change the topic frequently to keep them engaged",
    "d": "Speak over them to assert dominance"
  },
  "answer": "a"
}

Categories

Category	Questions
interpersonal_communication	8
marketing_and_sales	7
negotiation_and_strategic_interaction	12
personal_daily_habits	17
presentation_and_expression	18
public_communication	8
Total	70

Option Format

Most questions have four answer options. A small number have three options.

Number of options	Questions
3	7
4	63

Intended Uses

This dataset can be used for:

• Multiple-choice QA evaluation.
• Testing models on practical decision-making and influence knowledge.
• Prompting experiments for answer selection and rationale generation.
• Lightweight benchmark construction for communication and persuasion scenarios.

Loading Example

import json

with open("DecisionQE_dataset.json", "r", encoding="utf-8") as f:
    data = json.load(f)

questions = data["questions"]
print(len(questions))
print(questions[0]["question"])

Citation

No formal citation is currently provided. If you use this dataset, cite the Hugging Face dataset repository.