meetmate | Lindy Le

MeetMate supports preference elicitation and decision making within the context of meeting scheduling.

Project Overview

Many decision support systems are dependent on the ability to accurately model user preferences. However, people’s preferemces are contextual and often change throughout the decision-making process. Moreover, traditional optimization techniques require full knowledge of the model’s objective and constraints.

To address these issues, we built a system that combined LLMs with constraint programming: LLMs facilitated natural communication between the user and the system, translating user preferences into constraint functions for the optimization solver. We studied this hybrid framework through the lens of meeting scheduling, a time-consuming daily activity faced by a multitude of information workers.

MeetMate supports preference elicitation through preference construction and preference incorporation.

Approach

To evaluate our framework, we conducted three studies:

A diary study to characterize contextual scheduling preferences
A quantitative evaluation of the system’s performance
A user study that used our system as a technology probe to elicit design insights

MeetMate combines LLMs and Constraint Programming to facilitate meeting scheduling.

Results

Our work highlights the potential for a hybrid LLM and optimization approach for iterative preference elicitation and design considerations for building systems that support human-system collaborative decision-making processes (Lawless et al., 2024).

References

2024

ACM TiiS

“I Want It That Way”: Enabling Interactive Decision Support Using Large Language Models and Constraint Programming

Connor Lawless, Jakob Schoeffer, Lindy Le, Kael Rowan, Shilad Sen, Cristina St. Hill, Jina Suh , and Bahareh Sarrafzadeh

ACM Trans. Interact. Intell. Syst., Sep 2024

Abs DOI HTML PDF

A critical factor in the success of many decision support systems is the accurate modeling of user preferences. Psychology research has demonstrated that users often develop their preferences during the elicitation process, highlighting the pivotal role of system-user interaction in developing personalized systems. This paper introduces a novel approach, combining Large Language Models (LLMs) with Constraint Programming to facilitate interactive decision support. We study this hybrid framework through the lens of meeting scheduling, a time-consuming daily activity faced by a multitude of information workers. We conduct three studies to evaluate the novel framework, including a diary study to characterize contextual scheduling preferences, a quantitative evaluation of the system’s performance, and a user study to elicit insights with a technology probe that encapsulates our framework. Our work highlights the potential for a hybrid LLM and optimization approach for iterative preference elicitation, and suggests design considerations for building systems that support human-system collaborative decision-making processes.