Prevalence of Expression Types in Legislative Text
summary
A textual analysis of the Old Age Security Act measures how often specific expression types appear to guide a Rules-as-Code language. The work uses manual annotation plus regular expressions and boolean logic to detect categories, and a GPT-3 probe that scores 90%+ on four categories from 20 examples. Results show frequent cross-references and overwhelmingly common time-period expressions, while defeasibility/exception patterns are comparatively rare; a small prototype analyzer exists but is not yet public.
AI-written summary · about
related talks
18mData Extraction via Semantic Regular Expression Synthesis
16mDesigning Types for R, Empirically
01h 06mData and Society
54mDesign Patterns vs Anti pattern in APL
20mApplying Computational Analysis to Textual Data from the Wild: a Feminist Perspective
59mData-driven program analysis
29mGrammars of Data: lessons from ~20 years of the tidyverse
30mTowards a High Level Linter for Data Science