Source-linked AI summary
Personalizing Session-based Recommendations with Hierarchical Recurrent Neural Networks
Massimo Quadrana, Alexandros Karatzoglou, Balázs Hidasi, Paolo Cremonesi
TL;DR
Session-based recommenders usually rely on current-session interactions, even though some domains expose user identifiers and usable cross-session history. The paper introduces a Hierarchical RNN that transfers evolving user representations into session-level recommendation, and reports superior performance over session-only RNNs and item-based baselines on two industry datasets.
Problem
Session-based recommendation often lacks user profiles, but when identifiers exist, the value and modeling of past-session behavior for next-session recommendations remains a central challenge.
Method
A Hierarchical RNN adds a user-level GRU that evolves across sessions and initializes, optionally also informing, a session-level GRU.
Results
Hierarchical RNNs significantly outperform session-based RNNs and item-based alternatives, with reported gains including up to 6.5% Recall and 2.3% MRR over RNN Concat with large networks.
Takeaways & Limitations
Cross-session dynamics can personalize session-based recommendations and establish state-of-the-art performance for returning users within the evaluated industry settings.
Takeaways & Limitations
The authors identify extending personalized session-based models to domains such as music, e-commerce, and online advertising as future work.
Abstract
from arXiv · showhide
Session-based recommendations are highly relevant in many modern on-line services (e.g. e-commerce, video streaming) and recommendation settings. Recently, Recurrent Neural Networks have been shown to perform very well in session-based settings. While in many session-based recommendation domains user identifiers are hard to come by, there are also domains in which user profiles are readily available. We propose a seamless way to personalize RNN models with cross-session information transfer and devise a Hierarchical RNN model that relays end evolves latent hidden states of the RNNs across user sessions. Results on two industry datasets show large improvements over the session-only RNNs.
1 INTRODUCTION
Session-based recommendation uses current-session interactions, but identifiable users create an opportunity to transfer information across sessions. The paper proposes a Hierarchical RNN for this personalization and reports improvements over session-only and item-based alternatives on two industry datasets.
- Sessions group user interactions within a time frame and typically serve a goal such as finding a restaurant or listening to music.
- RNNs recently improved session-based recommendation over item-based methods by 15% to 30% in ranking metrics.
- When user identifiers are available, behavior from past sessions may provide valuable information for recommendations in a subsequent session.
- Concatenating past and current sessions is a simple personalization strategy, but it does not yield the best results experimentally.
- The proposed Hierarchical RNN transfers information across sessions while supporting both identifiable-user and no-identifier recommendation settings.
- On two industry datasets, Hierarchical RNNs outperform plain session-based RNNs and item-based collaborative filtering by a healthy margin.
2 RELATED WORK
Prior work includes item-to-item recommendation for anonymous sessions and RNNs for modeling sequential clicks. This paper extends session-based RNN modeling to user behavior across sessions.
- Session-based recommendations: Classical collaborative filtering can fail when session-based settings provide no user profile, motivating item-to-item recommendation from co-clicked session items.
- Session-based recommendations: Item-to-item methods precompute item similarities from session data and use frequently co-clicked items to generate recommendations.
- Recurrent Neural Models: RNNs are widely used for sequential data, with LSTMs adding gates that regulate input and hidden-state resets.
- Recurrent Neural Models: Earlier session RNNs used clicked item IDs from the current session to recommend the next item, whereas this work models behavior across sessions.
3 MODEL
The model adds a user-level GRU above a session-level GRU to evolve user representations across sessions and personalize next-session recommendations. Two variants differ in whether the user representation is only used for initialization or also propagated during the session.
- 3.1 Session-based Recurrent Neural Network: The baseline session RNN uses one GRU to model within-session interactions and output a score for every catalog item as the next-item likelihood.
- 3.1 Session-based Recurrent Neural Network: The session RNN processes current item IDs sequentially, initializes its hidden state with the null vector, and generates session-level representations.
- 3.1 Session-based Recurrent Neural Network: TOP1 loss is used because it outperformed the other tested ranking losses; it regularizes the relative rank of the relevant item using sampled irrelevant items.
- 3.2 Hierarchical RNN: HRNN adds a user-level GRU that tracks cross-session activity and user-interest evolution, alongside user-parallel mini-batches for efficient training.
- 3.2.1 Architecture: At session end, the user representation is updated from the session-level state; at the next session, it initializes the session GRU and may also enter each input.
- 3.2.1 Architecture: The user-level GRU learns session evolution rather than simply copying the previous session state, producing a user profile that personalizes recommendations.
- 3.2.1 Architecture: HRNN Init uses the user representation only to initialize the next session, while HRNN All additionally propagates it at every next-session step.
- 3.2.2 Learning: User-parallel batches group and time-order sessions by user, enabling training across users with different session counts and lengths.
4 EXPERIMENTS
Experiments compare personalized and session-based recommenders on XING and VIDEO, using sequential next-item prediction and Recall@5, MRR@5, and Precision@5. HRNNs generally outperform session-only RNNs and naïve concatenation, with gains varying by dataset, user-history length, and session position.
- Experimental setup: The evaluation uses sequential next-item prediction, bootstrapping HRNN and RNN Concat models with preceding user sessions in original order.Metrics assess the rank of the selected next item after each event.
- Results on XING: XING exhibits strong activity repetitiveness, making personalized recommendation difficult and making personalized popularity highly competitive.The session-based RNN is significantly worse than PPOP across network capacities.
- Results on XING: On XING, HRNNs significantly outperform RNN Concat in Recall and Precision and improve large-network MRR, while exceeding PPOP by approximately 11% in Recall and Precision with comparable MRR.HRNN gains over RNN Concat reach +3%/+1% in Recall for small/large networks and +5.4% MRR with HRNN All in large networks.
- Results on XING: On XING, HRNN All and HRNN Init have no significant differences in Recall or Precision, while HRNN All has a small approximately 2% MRR advantage.The comparison motivates further analysis on VIDEO.
- Results on VIDEO: On VIDEO, session-based RNN outperforms both Item-KNN and PPOP, while HRNN Init significantly outperforms all baselines and RNN Concat.In the reported large-network setting, HRNN Init is up to 6.5% better in Recall and 2.3% in MRR than RNN Concat.
- Analysis on user history length: Longer user histories improve Recall for all VIDEO methods, but improve MRR only for RNN Concat and HRNN Init; HRNN Init’s advantage over session-based RNN grows with history length.The Recall@5/MRR@5 gain grows from 5%/12% for short histories to 7%/19% for long histories.
- Analysis within sessions: Within VIDEO sessions, HRNN Init’s early-session advantage reaches 10% in Recall and 25% in MRR, while later session dynamics reduce Recall gains but not MRR gains.Personalization remains associated with higher ranking quality throughout the session.
- Analysis within sessions: On the small VIDEO dataset, HRNN Init outperforms session-based RNN by approximately 28% in Recall@5 and 41% in MRR@5.RNN Concat performs similarly to the session-based RNN in this evaluation.
5 CONCLUSIONS AND FUTURE WORK
The paper concludes that Hierarchical RNNs personalize session-based recommendation by transferring evolving user-interest representations across sessions, with HRNN Init performing best. It identifies further refinement through attention and item/user features, and broader evaluation across domains as future work.
- Hierarchical RNNs transfer long-term user-interest knowledge to session-level representations for personalized recommendations to returning users.The model adds a GRU level that models user activity across sessions and interest evolution over time.
- HRNNs significantly outperform state-of-the-art session-based RNNs and basic personalization strategies on two real-world datasets.
- HRNN Init, which initializes the session-level representation with the evolving user representation, gives the best results.
- The proposed approach sets new state-of-the-art performance for personalized session-based recommendation.
- Future work will examine attention models, item and user features, and personalized session-based models in domains such as music, e-commerce, and online advertising.