license: apache-2.0
task_categories:

• text-classification
• text-retrieval
  language:
• en
  tags:
• operations-research
• optimization
• mathematical-optimization
• linear-programming
• integer-programming
• reasoning
  pretty_name: Operations Research Question Classification
  size_categories:
• 1K<n<10K

Operations Research Question Classification Dataset

This dataset contains operations research (OR) questions classified into different categories with detailed reasoning.

Categories

The dataset includes questions classified into the following categories:

• Linear Programming (LP): Both objective function and constraints are linear, decision variables are continuous
• Integer Programming (IP): Both objective function and constraints are linear, all decision variables must be integers
• Mixed Integer Programming (MIP): Both objective function and constraints are linear, some decision variables are integers and some are continuous
• Nonlinear Programming (NP): At least one of the objective function or constraints is nonlinear
• Combinatorial Optimization (CO): Optimization problems involving discrete structures, such as routing, assignment, scheduling, etc.
• Multi-Objective Programming (MOP): Problems requiring simultaneous optimization of multiple objective functions

Dataset Statistics

• Total Questions: 2054
• Files: 6

Classification Distribution

• ****: 12 (0.6%)
• Combinatorial Optimization (CO): 53 (2.6%)
• Integer Programming (IP): 1046 (50.9%)
• Linear Programming (LP): 448 (21.8%)
• Mixed Integer Programming (MIP): 242 (11.8%)
• Nonlinear Programming (NP): 253 (12.3%)

Files

• MAMO_ComplexLP_fixed_classified.jsonl: 203 questions
• MAMO_EasyLP_fixed_classified.jsonl: 642 questions
• NL4OPT_classified.jsonl: 245 questions
• OptMATH_Bench_166_classified.jsonl: 166 questions
• OptMATH_Bench_193_classified.jsonl: 193 questions
• OptiBench_classified.jsonl: 605 questions

Dataset Structure

Each JSONL file contains questions with the following fields:

• en_question: The problem statement in English
• en_answer: The answer to the problem
• classification: The OR category (LP, IP, MIP, NP, CO, or MOP)
• reasoning: Detailed reasoning and analysis of why the problem belongs to the classified category
• classification_success: Whether the classification was successful

Usage

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("YOUR_USERNAME/OR-Question-Classification")

# Access the data
for example in dataset['train']:
    print(example['en_question'])
    print(example['classification'])
    print(example['reasoning'])

Source Files

This dataset is classified from the following sources:

• MAMO_ComplexLP_fixed: Complex linear programming problems
• MAMO_EasyLP_fixed: Easy linear programming problems
• NL4OPT: Natural language for optimization problems
• OptMATH_Bench_166: Optimization math benchmark
• OptMATH_Bench_193: Optimization math benchmark
• OptiBench: Optimization benchmark problems

Classification Methodology

Each problem was analyzed using an AI assistant with expertise in operations research. The classification process involved:

1. Analyzing decision variables
2. Examining the objective function (linear vs nonlinear)
3. Evaluating constraints (linear vs nonlinear)
4. Checking for integer requirements
5. Identifying multiple objectives (if any)
6. Assigning the most appropriate category

License

Apache 2.0

Citation

If you use this dataset, please cite:

@dataset{{or_question_classification,
  title={{Operations Research Question Classification Dataset}},
  author={{Your Name}},
  year={{2024}},
  url={{https://modelscope.cn/datasets/YOUR_USERNAME/OR-Question-Classification}}
}}