Standing on AI Giants: How InteraSSort Builds on Marketing and Tool Integration Research

Table of Links

Abstract and 1 Introduction

2. Related Work

3. Preliminaries

4. The InteraSSort Framework

5. Illustration

6. Conclusion and References

2. Related Work

In this study, we extend upon two key streams of research: (a) AI applications in marketing and (b) Language model tools. We briefly discuss some of the related works below.

Applications of AI in Marketing. Verma et al. (2021) explored the role of AI and disruptive technologies in business operations, explicitly highlighting the use of chatbots and language models to enhance the customer experience and customer relationship management systems. Similarly, De Mauro et al. (2022) presented a comprehensive taxonomy of machine learning and AI applications in marketing, emphasizing customer-facing improvements such as personalization, communication, recommendations, and assortments, as well as the benefits of machine learning on the business side, including market understanding and customer sense. In their literature review, Duarte et al. (2022) identified recommender systems and text analysis as promising areas for chatbot utilization in marketing. Fraiwan & Khasawneh (2023) discussed the applications, limitations, and future research directions pertaining to advanced language models in marketing. Building on earlier works, we explore the application of AI to solve assortment planning problem using LLMs.

Tools and their integration with LLMs. : Researchers have made significant strides in using LLMs to tackle complex tasks by extending their capabilities to include planning and API selection for tool utilization. For instance, Schick et al. (2023) introduced the pioneering work of incorporating external API tags into text sequences, enabling LLMs to access external tools. TaskMatrix.AI Liang et al. (2023) utilizes LLMs to generate high-level solution outlines tailored to specific tasks, matching subtasks with suitable off-the-shelf models or systems. HuggingGPT Shen et al. (2023) harnesses LLMs as controllers to effectively manage existing domain models for intricate tasks. Lastly, Qin et al. (2023) proposed a tool-augmented LLM framework that dynamically adjusts execution plans, empowering LLMs to proficiently complete each subtask using appropriate tools. Li et al. (2023b) introduced the Optiguide framework, leveraging LLMs to elucidate supply chain optimization solutions and address what-if scenarios. In contrast to the aforementioned approaches, InteraSSort harnesses the power of LLMs to enable interactive optimization in the context of assortment planning.