Yi Han

Real-world financial analysis involves information across multiple languages and modalities, from reports and news to scanned filings and meeting recordings. Yet most existing evaluations of LLMs in finance remain text-only, monolingual, and largely saturated by current models. To bridge these ga...

Most recommendation benchmarks evaluate how well a model imitates user behavior. In financial advisory, however, observed actions can be noisy or short-sighted under market volatility and may conflict with a user's long-term goals. Treating what users chose as the sole ground truth, therefore, co...

Click-through rate (CTR) prediction is a central task in recommender and information retrieval (IR) systems, with studies primarily benchmarking on AUC and LogLoss. However, when combined with other downstream components in a larger optimization pipeline to generate final ranking scores (e.g., ad...