A Universal Sets-level Optimization Framework for Next Set Recommendation

Yuli Liu,Min Liu,Christian Walder,Lexing Xie

2024-10-30

Abstract:Next Set Recommendation (NSRec), encompassing related tasks such as next basket recommendation and temporal sets prediction, stands as a trending research topic. Although numerous attempts have been made on this topic, there are certain drawbacks: (i) Existing studies are still confined to utilizing objective functions commonly found in Next Item Recommendation (NIRec), such as binary cross entropy and BPR, which are calculated based on individual item comparisons; (ii) They place emphasis on building sophisticated learning models to capture intricate dependency relationships across sequential sets, but frequently overlook pivotal dependency in their objective functions; (iii) Diversity factor within sequential sets is frequently overlooked. In this research, we endeavor to unveil a universal and S ets-level optimization framework for N ext Set Recommendation (SNSRec), offering a holistic fusion of diversity distribution and intricate dependency relationships within temporal sets. To realize this, the following contributions are made: (i) We directly model the temporal set in a sequence as a cohesive entity, leveraging the Structured Determinantal Point Process (SDPP), wherein the probabilistic DPP distribution prioritizes collections of structures (sequential sets) instead of individual items; (ii) We introduce a co-occurrence representation to discern and acknowledge the importance of different sets; (iii) We propose a sets-level optimization criterion, which integrates the diversity distribution and dependency relations across the entire sequence of sets, guiding the model to recommend relevant and diversified set. Extensive experiments on real-world datasets show that our approach consistently outperforms previous methods on both relevance and diversity.

Information Retrieval

What problem does this paper attempt to address?

### What problems does this paper attempt to solve?

This paper aims to solve several key problems in **Next Set Recommendation (NSRec)**. NSRec involves predicting the next set based on the user's past set sequences, such as the user's next purchase basket or a new playlist recommendation based on multi - day music listening patterns. Although much research has been devoted to this area, the following deficiencies still exist:

1. **Most existing studies use objective functions for Next Item Recommendation (NIRec)**: such as binary cross - entropy and Bayesian Personalized Ranking (BPR). These objective functions are based on single - item comparisons and fail to fully consider the complex dependencies within and between sets.

2. **The diversity factor within sets is ignored**: Many studies focus on constructing complex models to capture the dependencies between time - series sets, but ignore the importance of these dependencies in the objective function.

3. **Lack of emphasis on diversity within sets**: Diversity is crucial for the user experience during the recommendation process, but existing methods often do not fully consider this.

To solve these problems, this paper proposes a **general set - level optimization framework (SNSRec)** with the following main contributions:

- **Directly model time sets as a whole entity**, using the Structured Determinantal Point Process (SDPP) to prioritize structured sets rather than individual items.

- **Introduce co - occurrence representation** to identify the importance of different sets, thereby better understanding the relationships between sets.

- **Propose set - level optimization criteria** that combine the diversity distribution and the dependencies of the entire set sequence to guide the model to recommend relevant and diverse sets.

Through the above methods, this paper provides a comprehensive solution aimed at improving the relevance and diversity of NSRec tasks. Experimental results show that this method is more consistent on real - world datasets than previous methods.