wiba 0.2.0

WIBA: What Is Being Argued? A Comprehensive Approach to Argument Mining

WIBA: What Is Being Argued?

WIBA is a comprehensive argument mining toolkit that helps you detect, analyze, and understand arguments in text. It provides a simple yet powerful interface to identify argumentative content, extract topics, analyze stance, and discover arguments in longer texts.

Installation

pip install wiba

Quick Start

from wiba import WIBA

# Initialize with your API token
analyzer = WIBA(api_token="your_api_token_here")

# Example text
text = "Climate change is real because global temperatures are rising."

# Detect if it's an argument
result = analyzer.detect(text)
print(f"Argument detected: {result.argument_prediction}")
print(f"Confidence: {result.confidence_score}")

Features

• Comprehensive Analysis: Full argument analysis in one call - detection, extraction, topic, stance, and argument type
• Argument Detection: Identify whether a text contains an argument
• Topic Extraction: Extract the main topic being argued about
• Stance Analysis: Determine the stance towards a specific topic
• Argument Discovery: Find argumentative segments in longer texts
• Batch Processing: Efficiently process multiple texts
• DataFrame Support: Native pandas DataFrame integration

Documentation

For detailed documentation and examples, visit wiba.dev.

Getting Started

1. Create an account at wiba.dev to get your API token
2. Install the package: pip install wiba
3. Initialize the client with your token
4. Start analyzing arguments!

Example Usage

Detect Arguments

# Single text
result = analyzer.detect("Climate change is real because temperatures are rising.")
print(result.argument_prediction)  # "Argument" or "NoArgument"
print(result.confidence_score)     # Confidence score between 0 and 1

# Multiple texts
texts = [
    "Climate change is real because temperatures are rising.",
    "This is just a simple statement without any argument."
]
results = analyzer.detect(texts)
for r in results:
    print(f"Text: {r.text}")
    print(f"Prediction: {r.argument_prediction}")

Extract Topics

result = analyzer.extract("Climate change is a serious issue because it affects our environment.")
print(result.topics)  # List of extracted topics

Analyze Stance

text = "We must take action on climate change because the evidence is overwhelming."
topic = "climate change"
result = analyzer.stance(text, topic)
print(f"Stance: {result.stance}")  # "Favor", "Against", or "NoArgument"

Comprehensive Analysis (New in v0.2.0)

# Get full analysis in one call
text = "Climate change requires immediate action to prevent catastrophic environmental damage."
result = analyzer.comprehensive(text)

print(f"Is Argument: {result.is_argument}")      # True/False
print(f"Confidence: {result.confidence}")         # 0.0-1.0
print(f"Claims: {result.claims}")                 # List of claims
print(f"Premises: {result.premises}")             # List of premises
print(f"Topic (fine): {result.topic_fine}")       # Specific topic
print(f"Topic (broad): {result.topic_broad}")     # Broad policy area
print(f"Stance (fine): {result.stance_fine}")     # Favor/Against/NoArgument
print(f"Stance (broad): {result.stance_broad}")   # Favor/Against/NoArgument
print(f"Argument Type: {result.argument_type}")   # Deductive/Inductive/etc.
print(f"Argument Scheme: {result.argument_scheme}")

# Batch processing
texts = ["Text 1...", "Text 2..."]
results = analyzer.comprehensive(texts)

Discover Arguments

text = """Climate change is a serious issue. Global temperatures are rising at an
unprecedented rate. This is causing extreme weather events. However, some argue
that natural climate cycles are responsible."""

results_df = analyzer.discover_arguments(
    text,
    window_size=2,  # Number of sentences per window
    step_size=1     # Number of sentences to move window
)
print(results_df[['text_segment', 'argument_prediction', 'argument_confidence']])

Citation

If you use WIBA in your research, please cite:

@misc{irani2024wibaarguedcomprehensiveapproach,
      title={WIBA: What Is Being Argued? A Comprehensive Approach to Argument Mining}, 
      author={Arman Irani and Ju Yeon Park and Kevin Esterling and Michalis Faloutsos},
      year={2024},
      eprint={2405.00828},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2405.00828}, 
}

License

This project is licensed under the MIT License - see the LICENSE file for details.