Leveraging LLMs and Heterogeneous Knowledge Graphs for Persona-Driven Session-Based Recommendation

2026-04-08

AI Summary

This work addresses the challenge of modeling long-term user personality in anonymous conversational recommendation scenarios, where the absence of explicit user identifiers hinders personalization under cold-start and data-sparse conditions. The authors propose a two-stage framework that first integrates heterogeneous knowledge graphs with item embeddings generated by large language models (LLMs) to infer latent user personality from session behaviors via an unsupervised approach based on HDGI. In the second stage, the derived personality representations are jointly incorporated with short-term intent into the session-based recommendation model, complemented by a reranking mechanism to harmonize long- and short-term preferences. To the best of the authors’ knowledge, this is the first method to leverage structured relational signals for personality modeling without explicit user identifiers, achieving significant performance gains over purely sequential recommendation baselines on the Amazon Books and Movies & TV datasets.

Abstract

Session-based recommendation systems (SBRS) aim to capture user's short-term intent from interaction sequences. However, the common assumption of anonymous sessions limits personalization, particularly under sparse or cold-start conditions. Recent advances in LLM-augmented recommendation have shown that LLMs can generate rich item representations, but modeling user personas with LLMs remains challenging due to anonymous sessions. In this work, we propose a persona-driven SBRS framework that explicitly models latent user personas inferred from a heterogeneous knowledge graph (KG) and integrates them into a data-driven recommendation pipeline.Our framework adopts a two-stage architecture consisting of personalized information extraction and personalized information utilization, inspired by recent chain-of-thought recommendation approaches. In the personalized information extraction stage, we construct a heterogeneous KG that integrates time-independent user-item, item-item, item-feature association, and metadata from DBpedia. We then learn latent user personas in an unsupervised manner using a Heterogeneous Deep Graph Infomax (HDGI) objective over a KG initialized with LLM-derived item embeddings. In the personalized information utilization stage, the learned persona representations together with LLM-derived item embeddings are incorporated into a modified architecture of data-driven SBRS to generate a candidate set of relevant items, followed by reranking using the base sequential model to emphasize short-term session intent. Unlike prior approaches that rely solely on sequence modeling or text-based user representations, our method grounds user persona modeling in structured relational signals derived from a KG. Experiments on Amazon Books and Amazon Movies & TV demonstrate that our approach consistently improves over sequential models with user embeddings derived using session history.

Problem

• session-based recommendation
• user persona
• cold-start
• personalization
• anonymous sessions

Innovation

• LLM-augmented recommendation
• heterogeneous knowledge graph
• user persona modeling
• session-based recommendation
• Heterogeneous Deep Graph Infomax