FACE: A General Framework for Mapping Collaborative Filtering Embeddings into LLM Tokens

Chao Wang (University of Science and Technology of China) · Yixin Song (University of Science and Technology of China) · Jinhui Ye (Hong Kong University of Science and Technology(Guangzhou))) · Chuan Qin (Computer Network Information Center Chinese Academy of Sciences) · Dazhong Shen (Nanjing University of Aeronautics and Astronautics) · Lingfeng Liu (University of Science and Technology of China) · Xiang Wang (National University of Singapore) · Yanyong Zhang (Rutgers University, Newark)

benchmark modelscollaborative filteringconcept-specific vectorscontrastive alignmentdisentangled projectioninterpretability studiesinterpretable frameworklatent embeddings model-agnostic performance improvementspre-trained capabilitiesquantized autoencoderrecommendation systems semantic alignment

Recently, large language models (LLMs) have been explored for integration with collaborative filtering (CF)-based recommendation systems, which are crucial for personalizing user experiences. However, a key challenge is that LLMs struggle to interpret the latent, non-semantic embeddings produced by CF approaches, limiting recommendation effectiveness and further applications. To address this, we propose FACE, a general interpretable framework that maps CF embeddings into pre-trained LLM tokens. Specifically, we introduce a disentangled projection module to decompose CF embeddings into concept-specific vectors, followed by a quantized autoencoder to convert continuous embeddings into LLM tokens (descriptors). Then, we design a contrastive alignment objective to ensure that the tokens align with corresponding textual signals. Hence, the model-agnostic FACE framework achieves semantic alignment without fine-tuning LLMs and enhances recommendation performance by leveraging their pre-trained capabilities. Empirical results on three real-world recommendation datasets demonstrate performance improvements in benchmark models, with interpretability studies confirming the interpretability of the descriptors. Code is available in \url{https://github.com/YixinRoll/FACE}.