Simulating Word Suggestion Usage in Mobile Typing to Guide Intelligent Text Entry Design

TL;DR

This paper introduces WSTypist, a reinforcement learning model that simulates human word suggestion usage in mobile typing, aiding design without extensive user studies.

Contribution

It extends hierarchical control models by incorporating cognitive mechanisms, enabling simulation of diverse suggestion strategies and user adaptation.

Findings

WSTypist reproduces individual differences in suggestion use.

The model generalizes across different text entry systems.

It can inform design choices through what-if analyses.

Abstract

Intelligent text entry (ITE) methods, such as word suggestions, are widely used in mobile typing, yet improving ITE systems is challenging because the cognitive mechanisms behind suggestion use remain poorly understood, and evaluating new systems often requires long-term user studies to account for behavioral adaptation. We present WSTypist, a reinforcement learning-based model that simulates how typists integrate word suggestions into typing. We extend recent hierarchical control models of typing, by identifying and implementing important cognitive mechanisms that underlie the high-level decision-making for integrating word suggestions into manual typing: considering orthographic processes, assessing efficiency gains, and including personal preference on AI support. Our evaluations show that WSTypist simulates diverse human-like suggestion-use strategies, reproduces individual differences, and generalizes across different systems. Importantly, we demonstrate on four design cases how a computational rationality model can be used to inform what-if analyses during the design process, by simulating how users might adapt to changes in the UI or in the algorithmic support, reducing the need for long-term user studies.