Útmutatóhoz kapcsolódó kvíz · Közepes Szint

AI Clause Extraction and Contract Metadata Quiz

Tests clause extraction, source traceability, interacting terms, metadata fields, validation, and limits of legal automation.

Kapcsolódó útmutató utakAI Clause Extraction And Contract Metadata
kérdés 1 -ból 8

An AI system fills in a contract’s renewal date. What should the reviewer verify?

Tekintse meg az összes kérdést8 kérdéseket

A válaszok és magyarázatok megjelennek a kvíz kitöltésekor.

  1. An AI system fills in a contract’s renewal date. What should the reviewer verify?
    • The source clause, any notice period, amendments, and linked termination terms
    • The renewal clause and notice window, without checking for a superseding amendment
    • The signature date, treated as renewal date unless a model flags a conflict
    • The date from another agreement with the same counterparty
  2. What does clause extraction produce most directly?
    • A complete clause summary and legal interpretation without retaining source text
    • Candidate text spans or structured fields linked to a contract
    • A normalized field value without the source span or document location
    • A recommendation about whether a party should accept or renegotiate the clause
  3. Why store the original clause passage with an extracted metadata field?
    • Store the normalized field only and discard the agreement version
    • Replace the executed contract with an extracted clause summary
    • Treat the first extracted value as final even after an amendment
    • So a reviewer can verify the value against the source and preserve provenance
  4. An indemnity provision has a broad opening sentence followed by exceptions. What should the workflow do?
    • Keep the grant language and disregard exceptions because they are secondary
    • Capture the exceptions and surrounding definitions for reviewer assessment
    • Classify the provision from its heading without tracing defined terms
    • Separate each carve-out from the clause and discard its cross-references
  5. A scanned amendment changes a date extracted from the original contract. Which risk should be tested?
    • OCR and document-version handling may cause the system to miss the amendment
    • Normalize the contract date using its signature date even if the effective date differs
    • Choose the latest date in the file without checking whether it changes the extracted field
    • Assume the extracted date remains correct because the amendment is a separate file
  6. Which evaluation best measures a contract field extractor?
    • Report throughput across one template family without measuring field errors
    • Compare confidence scores with the model’s own predicted answer
    • Evaluate one agreement type and assume performance transfers across contract classes
    • Test span and field accuracy, false negatives, false positives, and reviewer corrections
  7. How should an extracted governing-law field be presented to users?
    • As a candidate linked to its source and review status
    • As a conclusive choice-of-law determination without review of amendments or related clauses
    • As a legal opinion that the contract will be enforced in that jurisdiction
    • As an inferred jurisdiction based only on a party’s address
  8. A contract repository uses AI-generated renewal reminders. What control is most important?
    • Use an extracted date without checking a superseding amendment or notice period
    • Have an authorized reviewer confirm dates and notice conditions before consequential alerts
    • Rely on confidence score to trigger notice before a person verifies the contract
    • Use a standard renewal term when the contract text is missing