Modeling Copilots for Text-to-Model Translation

TL;DR

This paper introduces Text2Model and Text2Zinc, a unified, solver-agnostic framework and dataset for translating natural language into combinatorial models, with experiments comparing various LLM strategies.

Contribution

It presents the first unified architecture and dataset for satisfaction and optimization problems, leveraging MiniZinc for solver-agnostic modeling, and provides open-source tools and benchmarks.

Findings

LLMs show promise but are not yet a push-button solution for combinatorial modeling.

Various LLM strategies, including chain-of-thought and knowledge-graphs, improve solution accuracy.

Open-source copilots and datasets are provided to advance research in this area.

Abstract

There is growing interest in leveraging large language models (LLMs) for text-to-model translation and optimization tasks. This paper aims to advance this line of research by introducing \textsc{Text2Model} and \textsc{Text2Zinc}. \textsc{Text2Model} is a suite of copilots based on several LLM strategies with varying complexity, along with an online leaderboard. \textsc{Text2Zinc} is a cross-domain dataset for capturing optimization and satisfaction problems specified in natural language, along with an interactive editor with built-in AI assistant. While there is an emerging literature on using LLMs for translating combinatorial problems into formal models, our work is the first attempt to integrate \textit{both} satisfaction and optimization problems within a \textit{unified architecture} and \textit{dataset}. Moreover, our approach is \textit{solver-agnostic} unlike existing work that focuses on translation to a solver-specific model. To achieve this, we leverage \textsc{MiniZinc}'s solver-and-paradigm-agnostic modeling capabilities to formulate combinatorial problems. We conduct comprehensive experiments to compare execution and solution accuracy across several single- and multi-call strategies, including; zero-shot prompting, chain-of-thought reasoning, intermediate representations via knowledge-graphs, grammar-based syntax encoding, and agentic approaches that decompose the model into sequential sub-tasks. Our copilot strategies are competitive, and in parts improve, recent research in this domain. Our findings indicate that while LLMs are promising they are not yet a push-button technology for combinatorial modeling. We contribute \textsc{Text2Model} copilots and leaderboard, and \textsc{Text2Zinc} and interactive editor to open-source to support closing this performance gap.