A Reproducibility Analysis of PO4ISR: Diagnosing and Mitigating Semantic Drift in LLM-Based Session Recommendation
The article presents a reproducibility analysis of the PO4ISR model for session-based recommendations. It identifies significant performance issues due to semantic drift in long sessions and proposes an enhanced version called PO4ISR++. The The study demonstrates that the new implementation improves performance across various datasets, confirming its robustness.
- ▪The original PO4ISR model struggles with contextual drift in long session recommendations.
- ▪The new implementation, PO4ISR++, incorporates reflexive prompting to enhance stability.
- ▪The enhanced model shows performance gains of up to 54% on the Games dataset and 96% on the Bundle dataset.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Information Retrieval arXiv:2605.18780 (cs) [Submitted on 29 Apr 2026] Title:A Reproducibility Analysis of PO4ISR: Diagnosing and Mitigating Semantic Drift in LLM-Based Session Recommendation Authors:Aditya Tiwari, Konduri Naga Lakshmi Rekha, Rajesh Kumar Mundotiya View a PDF of the paper titled A Reproducibility Analysis of PO4ISR: Diagnosing and Mitigating Semantic Drift in LLM-Based Session Recommendation, by Aditya Tiwari and Konduri Naga Lakshmi Rekha and Rajesh Kumar Mundotiya View PDF HTML (experimental) Abstract:Reasoning-based Large Language Models (LLMs) like PO4ISR have set new benchmarks in session-based recommendation. However, the reproducibility of their reasoning capabilities across diverse semantic domains remains unexplored.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
