Loading...

Media is loading
 

Paper Number

ICIS2025-1241

Paper Type

Short

Abstract

Autonomous web agents are increasingly capable of completing real-world tasks, yet they struggle to stay aligned with evolving user intent and dynamic web layouts. We present Web2HMM, a Hidden Markov Model-based framework that enables real-time human-in-the-loop interaction during agent execution. By inferring latent decision states, AUTO for confident actions and PAUSE for seeking user input, the agent dynamically balances autonomous operation with timely, proactive feedback. Applied to simulated hotel booking scenarios, Web2HMM demonstrates more adaptive and user-aware behavior, advancing the design of collaborative and resilient web agents.

Comments

15-Interaction

Share

COinS
 
Dec 14th, 12:00 AM

Toward Interactive Web Agents: Proactive Feedback through Hidden Markov-Based Decision Modeling

Autonomous web agents are increasingly capable of completing real-world tasks, yet they struggle to stay aligned with evolving user intent and dynamic web layouts. We present Web2HMM, a Hidden Markov Model-based framework that enables real-time human-in-the-loop interaction during agent execution. By inferring latent decision states, AUTO for confident actions and PAUSE for seeking user input, the agent dynamically balances autonomous operation with timely, proactive feedback. Applied to simulated hotel booking scenarios, Web2HMM demonstrates more adaptive and user-aware behavior, advancing the design of collaborative and resilient web agents.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.