IMPACT-P: New Approach for Bundle Recommendation Using Implicit Constraints and Sentiment-Based Reranking

• Paper Link: Journal of Information Processing Systems (JIPS)
• Code & Dataset Repository: Hugging Face - IMPACT-P
• DOI: 10.3745/JIPS.04.0377

Abstract

Constraint-based bundle recommendations are typically derived from explicit user requirements, often resulting in bundles with low or even nonexistent diversity. Consequently, recent studies have shifted their focus toward enhancing diversity in recommendations. However, this approach tends to compromise recommendation relevance. To address this issue, this study proposes a novel framework called IMPACT-P (implicit matching and constraint tuning for personalization), which leverages implicit user preferences and matrix sentiment-based reranking of bundles. The proposed framework significantly improves recommendation relevance, achieving up to a 300% increase in hit ratio, while diversity decreases by only 0.6% compared to the baseline framework (MF_ALL).

Citation

APA Style

Yusuf, A., Adiwijaya, Wibowo, A. T., & Baizal, Z. K. A. (2026). IMPACT-P: New Approach for Bundle Recommendation Using Implicit Constraints and Sentiment-Based Reranking. Journal of Information Processing Systems, 22(4), 333-347. DOI: 10.3745/JIPS.04.0377.

IEEE Style

A. M. Yusuf, Adiwijaya, A. T. Wibowo, and Z. K. A. Baizal, "IMPACT-P: New Approach for Bundle Recommendation Using Implicit Constraints and Sentiment-Based Reranking," Journal of Information Processing Systems, vol. 22, no. 4, pp. 333-347, Aug. 2026. DOI: 10.3745/JIPS.04.0377.

ACM Style

Andy Maulana Yusuf, Adiwijaya, Agung Toto Wibowo, and Z. K. A. Baizal. 2026. IMPACT-P: New Approach for Bundle Recommendation Using Implicit Constraints and Sentiment-Based Reranking. Journal of Information Processing Systems, 22, 4 (Aug. 2026), 333–347. DOI: 10.3745/JIPS.04.0377.