Reproduce “AutoQual: An LLM Agent for Automated Discovery of Interpretable Features for Review Quality Assessment” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xiaochong Lan, Jie Feng, Yinxing Liu, Xinlei Shi, Yong Li
AutoQual is an LLM agent framework that autonomously identifies explicit and computable features for assessing online review quality, successfully translating tacit expert knowledge from labeled data into interpretable insights. The system, developed by researchers from Tsinghua University and Meituan, demonstrated improved predictive performance when combined with PLM embeddings and led to a 0.27% increase in review reader conversion rates during industrial deployment on Meituan's platform.
Key insights
- The structured agent design, including multi-perspective ideation and contrastive analysis, is crucial for successful feature discovery, as direct LLM baselines performed poorly.
- AutoQual's interpretable features provide valuable, complementary information to PLM embeddings, consistently achieving top performance when combined.
- A dual-level memory system, with intra-task reflection and cross-task knowledge transfer, significantly enhances performance and computational efficiency.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.