Contract Understanding Atticus Dataset (CUAD) Subtask

A Contract Understanding Atticus Dataset (CUAD) Subtask is a span-selection legal contract analysis task that can support CUAD contract review tasks.

• AKA: CUAD Subtask.
• Context:
  • It can typically extract CUAD Text Spans corresponding to CUAD legal clause categories through CUAD token position prediction.
  • It can typically identify CUAD Clause Presence through CUAD binary classification for CUAD 33 clause types.
  • It can typically extract CUAD Specific Entities including CUAD temporal entities, CUAD geographical entities, and CUAD numerical values.
  • It can typically recognize CUAD Multiple Text Spans within CUAD single contracts for CUAD complex clauses.
  • It can typically process CUAD Contract Documents through CUAD sliding window techniques for CUAD lengthy contracts.
  • ...
  • It can often utilize CUAD Evaluation Metrics including CUAD exact match, CUAD F1 score, and CUAD AUPR metric.
  • It can often employ CUAD Jaccard Similarity for CUAD span matching with CUAD 0.5 threshold.
  • It can often address CUAD Class Imbalance where CUAD relevant clauses comprise only CUAD 10% of contract content.
  • It can often require CUAD Multi-Annotator Verification for CUAD ground truth labeling.
  • ...
  • It can range from being a Simple CUAD Clause Detection Subtask to being a Complex CUAD Entity Extraction Subtask, depending on its CUAD extraction complexity.
  • It can range from being a Binary CUAD Classification Subtask to being a Multi-Label CUAD Classification Subtask, depending on its CUAD label structure.
  • It can range from being a Clause-Level CUAD Analysis Subtask to being a Document-Level CUAD Analysis Subtask, depending on its CUAD analysis granularity.
  • ...
  • It can measure CUAD Precision at 80% Recall for CUAD performance evaluation.
  • It can measure CUAD Precision at 90% Recall for CUAD stringent assessment.
  • It can generate CUAD Confidence Probabilities for CUAD threshold-based evaluation.
  • It can perform CUAD Category-Specific Analysis across CUAD 41 clause categories.
  • It can apply CUAD Downweighting Strategies for CUAD training balance.
  • ...
• Example(s):
  • CUAD Simple Clause Detection Subtasks, such as:
    • CUAD Anti-Assignment Clause Detection for CUAD Yes/No determination.
    • CUAD Non-Compete Provision Detection for CUAD binary classification.
    • CUAD Governing Law Specification Detection for CUAD clause presence identification.
  • CUAD Complex Entity Extraction Subtasks, such as:
    • CUAD Temporal Entity Extractions, such as:
      • CUAD Effective Date Extraction for CUAD contract start date.
      • CUAD Expiration Date Extraction for CUAD contract end date.
      • CUAD Renewal Term Extraction for CUAD contract extension period.
    • CUAD Geographical Entity Extractions, such as:
      • CUAD Governing Jurisdiction Extraction for CUAD legal jurisdiction.
      • CUAD Territorial Restriction Extraction for CUAD geographic limitation.
    • CUAD Numerical Value Extractions, such as:
      • CUAD Liability Cap Extraction for CUAD damage limitation.
      • CUAD Minimum Commitment Extraction for CUAD obligation threshold.
      • CUAD Revenue Percentage Extraction for CUAD financial term.
    • CUAD Named Entity Extractions, such as:
      • CUAD Party Name Extraction for CUAD contracting parties.
      • CUAD Document Title Extraction for CUAD contract name.
  • CUAD Multi-Span Recognition Subtasks, such as:
    • CUAD Multiple Instance Clause Extraction for CUAD repeated clause type.
    • CUAD Non-Contiguous Text Extraction for CUAD distributed clause information.
  • CUAD High-Performance Category Subtasks, such as:
    • CUAD Document Name Extraction achieving CUAD high AUPR score.
    • CUAD Governing Law Extraction demonstrating CUAD strong performance.
  • CUAD Complex Category Subtasks, such as:
    • CUAD Covenant Not to Sue Extraction showing CUAD lower AUPR score.
    • CUAD Most Favored Nation Clause Extraction with CUAD recognition challenge.
  • ...
• Counter-Example(s):
  • General Question Answering Task, which lacks CUAD legal domain specificity and CUAD contract focus.
  • Legal Case Prediction Task, which predicts legal outcomes rather than extracting CUAD contract clauses.
  • Document Summarization Task, which generates document summaries rather than identifying CUAD specific clause spans.
  • Named Entity Recognition Task, which extracts general entities without CUAD legal contract context.
  • Text Classification Task, which categorizes entire documents rather than extracting CUAD precise text spans.
• See: Contract Understanding Atticus Dataset (CUAD) Benchmark, Contract Understanding Atticus Dataset (CUAD) Clause Label, Legal Contract Analysis Task, Span Selection Question Answering Task, Legal NLP Evaluation Metric, Contract Clause Extraction Task, Legal Entity Extraction Task.

---