Uncovering Global Localization Strategies for Generative AI Applications: An Analysis of 2023-2025 iOS Big Data via Panel Regression and Neural Topic Modeling

Key Points

• This research aims to analyze the global market performance drivers and localization strategies of major generative AI applications from 2023 to 2025.
• Conducted panel regression analysis with fixed effects to control for country-specific characteristics.
• Utilized advanced neural topic modeling combining Llama 3.1 and Mistral-Nemo models.
• Integrated quantitative performance data with qualitative user responses.
• Categorized global markets based on user engagement patterns.
• User evaluation metrics positively affect downloads and revenue.
• Observed the Rating Paradox, where increased user complaints coexist with positive performance metrics.
• Identified four global market types: signal-sensitive, qualitative value-selective, disruptive/mature paradoxical, and market-inertial.
• Classified user interests into six domains highlighting emotional bonding and infrastructure stability as core factors.

Abstract