AutoRecLab: Describe the Experiment, Get the Code!
AutoRecLab turns a natural-language RecSys experiment idea into runnable Python code, then checks and expands it through execution feedback.
The paper presents it as an autonomous lab for recommender-systems experiments, using RAG for documentation lookup, static type checks, and execution-steered tree search. In the demo, it implements an explicit-to-implicit feedback conversion study. The authors report that, in a baseline comparison across six algorithms and three datasets, 8 of 9 runs succeeded at about $1 per run using GPT-5.4-mini. The paper was accepted to the RecSys ’26 Demo Track. ArXiv · AI/CL/LG's note
The paper presents it as an autonomous lab for recommender-systems experiments, using RAG for documentation lookup, static type checks, and execution-steered tree search. In the demo, it implements an explicit-to-implicit feedback conversion study. The authors report that, in a baseline comparison across six algorithms and three datasets, 8 of 9 runs succeeded at about $1 per run using GPT-5.4-mini. The paper was accepted to the RecSys ’26 Demo Track. ArXiv · AI/CL/LG's note
score 5