Instruction-aware User Embedding via Synergistic Language and Representation Modeling

2025-10-13

AI Summary

Existing user representation methods suffer from poor cross-domain generalization and vulnerability to behavioral noise. To address these limitations, we propose the Instruction-aware User Embedding Model (IUEM), a foundation model for user embeddings that leverages large language models (LLMs) to generate representations with both strong generalizability and instruction-following capability. Methodologically, IUEM employs a multi-encoder architecture augmented with lightweight adapters to jointly encode linguistic and embedding spaces; it further introduces a contrastive–autoregressive joint training framework, enabling instruction-guided behavioral denoising and robust cross-domain modeling. The model is co-optimized on the UserQA dataset. Extensive experiments demonstrate that IUEM significantly outperforms state-of-the-art methods across real-world tasks—including user prediction, marketing, and recommendation—validating its superior cross-domain adaptability and robustness to behavioral noise.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Humans and AI: Intelligent User Interfaces
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

User representation modeling has become increasingly crucial for personalized applications, yet existing approaches struggle with generalizability across domains and sensitivity to noisy behavioral signals. We present InstructUE, an instruction-aware user embedding foundation model that leverages large language models (LLMs) to generate general and instruction-aware user representations. InstructUE introduces a multi-encoder architecture with a lightweight adapter that efficiently processes heterogeneous data from six different sources while preserving their structural characteristics. Additionally, it proposes a novel contrastive-autoregressive training framework that bridges language and representation spaces through a curated UserQA dataset. The contrastive-autoregressive training framework simultaneously leverages autoregressive learning to capture domain knowledge in language space and contrastive learning to align user-text embeddings in representation space, thereby enhancing the instruction-awareness and noise-robustness of user embeddings. Through extensive experiments on real-world applications, we demonstrate that InstructUE significantly outperforms existing methods across multiple domains including user prediction, marketing, and recommendation scenarios. Our results show that instruction-aware user modeling can effectively achieve instruction-guided denoising of user information in specific scenarios, paving the way for more generalizable and robust user representation learning.

Problem

• Creating generalizable user embeddings across domains with noisy behavioral data
• Aligning language and representation spaces for instruction-aware modeling
• Developing noise-robust user representations through multi-source heterogeneous data processing

Innovation

• Leverages LLMs for instruction-aware user embeddings
• Uses multi-encoder architecture with lightweight adapter
• Proposes contrastive-autoregressive training framework