Researchers have introduced SocialPersona, a new benchmark designed to evaluate the ability of multimodal large language models (MLLMs) to infer user preferences from social media data. The benchmark utilizes longitudinal timelines from 171 social media users, incorporating text, images, and timestamps, along with human-verified preference tags. SocialPersona supports tasks such as constructing user profiles and generating personalized responses, with experiments indicating that while MLLMs can identify broad interests, they struggle with fine-grained and recent preferences, highlighting a key challenge in cross-modal user modeling. AI
IMPACT This benchmark aims to advance the development of AI assistants that can infer and act on user preferences, potentially leading to more personalized and effective AI interactions.
RANK_REASON The cluster describes a new academic benchmark for evaluating multimodal large language models.
Read on arXiv cs.IR (Information Retrieval) →
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
