Papers by David Stein
Building a Long Text Privacy Policy Corpus with Multi-Class Labels (2025.acl-long)
Copied to clipboard
| Challenge: | Legal text is susceptible to multiple valid, conflicting interpretations, and indeterminacy, interdependence between clauses, meaningful silence, and implications of legal defaults. |
| Approach: | They propose to annotate privacy policies from 149 firms using a hand-coded dataset that captures key challenges peculiar to legal language. |
| Outcome: | The proposed dataset includes privacy policies from 149 firms and includes materials incorporated by reference. |