A Multi-Stage Workflow for the Review of Marketing Content with Reasoning Large Language Models
This addresses the specific problem of marketing content compliance review for businesses, representing an incremental application of existing LLM methods to a new domain.
The paper tackles the problem of automating compliance review for marketing content by proposing a multi-stage workflow using fine-tuned reasoning LLMs, achieving results that show the effectiveness of different fine-tuning strategies and reward functions.
Reasoning Large Language Models (LLMs) have shown promising results when tasked with solving complex problems. In this paper, we propose and evaluate a multi-stage workflow that leverages the capabilities of fine-tuned reasoning LLMs to assist in the review process of marketing content, making sure they comply with a given list of requirements. The contributions of this paper are the following: (i) we present a novel approach -- that does not rely on any external knowledge representation -- for the automatic identification of compliance issues in textual content; (ii) compare the effectiveness of different fine-tuning strategies like Supervised Fine-Tuning (SFT) and Group Relative Policy Optimization (GRPO) in training models to solve this problem; (iii) we evaluate the effectiveness of training small LLMs to generate reasoning tokens before providing their final response; (iv) we evaluate how the choice and combinations of different reward functions affects the performance of a model trained with GRPO.