Symbolic Prompt Compression Improves Boolean Evaluation: Evidence from Z3-Verified Logical Compression

Read full text Summarise with VIRAT
Authors: Emil Piotr Dobkowski
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Z3-verified symbolic compression (Recursive SDC v2.1) improves boolean evaluation accuracy by 24% over natural language prompts (Acc_SDC=0.880 vs Acc_NL=0.640, delta=-0.240, CI=[-0.400,-0.080], n=25, DeepSeek-V3, SEED=42). The hypothesized cognitive collapse at CR>3.4 (char-based) is unattainable under semantic-preserving compression: safe ceiling is CR_tok=1.38 for propositional logic, above which Z3 detects semantic violation (equiv=False). Character-based CR overestimates token-based CR by +62.9% (n=8, tiktoken cl100k_base), making prior char-CR thresholds unreliable. SDC improves accuracy by eliminating NL parsing overhead, leaving only formal logical structure for evaluation.

Fields of Study & Subjects

Formal Methods in Verification Topic Modeling Natural Language Processing Techniques

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
22 March 2026
Language
EN
Document type
Preprint