Researchers have developed TaxoScale, a new pipeline designed to automatically generate taxonomies of security and privacy concerns from mobile app reviews. This method addresses the scalability challenge faced by existing LLM and clustering techniques, which are typically limited to smaller datasets. TaxoScale processes a corpus of over 600,000 app reviews, extending expert-defined taxonomies through recursive hierarchical clustering and LLM-based naming to discover novel branches and outperform baseline methods. AI
IMPACT This research could enable more efficient and scalable identification of emerging security and privacy risks within software ecosystems.
RANK_REASON The cluster describes a new research paper detailing a novel algorithm and pipeline for automated taxonomy generation from a large dataset.
Read on Hugging Face Daily Papers →
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
