14 projects
JAAT
Repository for JAAT: efficient and accurate analysis of job ads for task/skill matching, title matching, firm name extraction, attribute classification, and wage extraction.
dstpr
A high-throughput, domain-specific text preprocessing cascading pipeline filter to rank core sentences in texts and filter away boilerplate.
dpmlm
DP-MLM: Differentially Private Text Rewriting Using Masked Language Models
similarity-tool
A modular, hybrid, and customizable document similarity framework.
subsegmenter
A lightweight semantic sub-sentence segmenter using rule-optimized spaCy dependency trees.
dpst
Differentially private document generation leveraging semantic triples.
llm-judge-panel
A multi-provider LLM-as-a-Judge consensus panel with chained criteria for concept matching validation.
jaat-api
Official lightweight client wrapper library for the JAAT API.
TaxonomyBuilder
A robust tool for the automated building of custom, domain-specific taxonomies using LLMs and GPU-accelerated clustering.
mldp-text
Word-level Metric Local Differential Privacy Mechanisms
privfill
LLM-based Differential Privacy mechanisms for sentence-based text rewriting with infilling models.
dp-diffractor
1-Diffractor: a highly efficient word-level Metric Differential Privacy mechanism.
CSKE
Efficient and powerful class-specific keyword extraction.
concat-ls
Repository for ConCat: a simple and intuitive method for English lexical substitution.
