Investigating the Potential of Using Large Language Models for Scheduling

TL;DR

This paper explores the use of Large Language Models for conference program scheduling, demonstrating their ability to generate initial drafts and cluster papers effectively using only titles, with results comparable to human categorization.

Contribution

It introduces a novel application of LLMs for scheduling and clustering in conference settings, emphasizing zero-shot learning and demonstrating their effectiveness with minimal input.

Findings

LLMs can produce good initial conference schedules in zero-shot settings.

Using only paper titles with LLMs yields clustering results closer to human categorization.

The study provides publicly available code for replication.

Abstract

The inaugural ACM International Conference on AI-powered Software introduced the AIware Challenge, prompting researchers to explore AI-driven tools for optimizing conference programs through constrained optimization. We investigate the use of Large Language Models (LLMs) for program scheduling, focusing on zero-shot learning and integer programming to measure paper similarity. Our study reveals that LLMs, even under zero-shot settings, create reasonably good first drafts of conference schedules. When clustering papers, using only titles as LLM inputs produces results closer to human categorization than using titles and abstracts with TFIDF. The code has been made publicly available.