Construction and Analysis of a Five-Factor Personality Assessment Model for Large Language Models (LLMs)

Thu, 01 May 2025 00:00:00 +0000

This project develops a multidimensional assessment framework for studying AI agent personality traits in large language models. Based on the Big Five model and personality-oriented prompts inspired by psychological scales such as NEO-PI-R, it quantitatively compares behavioral differences among models including DeepSeek-V3 and Qwen 2.5.

The analysis uses Python to process model-generated text and statistical methods, including correlation analysis, to compare model behavior across personality dimensions. The project provides benchmark-style results for future research on AI agent behavioral consistency and personality modeling.

Financial Time Series Volatility Breakpoint Detection under a Bayesian Framework

Wed, 01 Jan 2025 00:00:00 +0000

This course project builds a Bayesian model for detecting structural changes in financial time series volatility. It derives a joint likelihood with a discrete breakpoint and distinct volatility parameters, then estimates the model with a Random Walk Metropolis-Hastings sampler.

The project compares Uniform, Inverse-Gamma, and Exponential prior settings and applies the method to 2008 S&P 500 index data. The results are used to interpret volatility changes around the Lehman Brothers bankruptcy.

Projects | YuxiaDing's homepage

Construction and Analysis of a Five-Factor Personality Assessment Model for Large Language Models (LLMs)

Financial Time Series Volatility Breakpoint Detection under a Bayesian Framework