A Change Talk Model for Abstinence Based on Web-Based Anonymous Gambler Chat Meeting Data by Using an Automatic...

Area of Science:

• Psychology
• Computer Science
• Data Science

Background:

• Gambling disorder is a significant public health issue.
• Understanding the process of gambling abstinence is crucial for effective interventions.
• Previous research has focused on talk patterns but not specifically on the abstinence process in anonymous online settings.

Purpose of the Study:

• To develop a computational model for analyzing change and sustain talk in anonymous gambler chat meetings.
• To create a model that can predict gambling abstinence based on linguistic patterns.
• To identify the optimal proportion of change talk for successful abstinence.

Main Methods:

• Utilized a large dataset of 1.63 million utterances from Japanese web-based anonymous gambler chat meetings.
• Employed an automatic change talk classifier to categorize utterances into 'change talk' and 'sustain talk'.
• Applied a support vector machine with a radial basis kernel function to build the abstinence prediction model.

Main Results:

• Abstinent gamblers exhibited significantly higher proportions of change talk and lower proportions of sustain talk compared to non-abstinent gamblers.
• The developed model accurately classified abstinent and non-abstinent gamblers.
• The model provided personalized evaluation scores and suggested optimal change talk proportions for individuals.

Conclusions:

• Personalized feedback on talk patterns can enhance gambling abstinence likelihood.
• This approach may help mitigate the severe mental, social, and financial consequences of gambling disorder.
• Leveraging computational analysis of online discussions offers a novel avenue for supporting gambling recovery.