Session 5: Cross-domain and Cross-modal Learning

Date: Tuesday October 15, 17:20 PM – 18:45 PM (GMT+2)
Room: Petruzzelli Theater
Session Chair: Robin Burke

• RES 🕓15Improving Adversarial Robustness for Recommendation Model via Cross-Domain Distributional Adversarial Training
  by Jingyu Chen (Sichuan University), Lilin Zhang (Sichuan University) and Ning Yang (Sichuan University)

  Recommendation models based on deep learning are fragile when facing adversarial examples (AE). Adversarial training (AT) is the existing mainstream method to promote the adversarial robustness of recommendation models. However, these AT methods often have two drawbacks. First, they may be ineffective due to the ubiquitous sparsity of interaction data. Second, point-wise perturbation used by these AT methods leads to suboptimal adversarial robustness, because not all examples are equally susceptible to such perturbations. To overcome these issues, we propose a novel method called Cross-domain Distributional Adversarial Training (CDAT) which utilizes a richer auxiliary domain to improve the adversarial robustness of a sparse target domain. CDAT comprises a Domain adversarial network (Dan) and a Cross-domain adversarial example generative network (Cdan). Dan learns a domain-invariant preference distribution which is obtained by aligning user embeddings from two domains and paves the way to leverage the knowledge from another domain for the target domain. Then, by adversarially perturbing the domain-invariant preference distribution under the guidance of a discriminator, Cdan captures an aggressive and imperceptible AE distribution. In this way, CDAT can transfer distributional adversarial robustness from the auxiliary domain to the target domain. The extensive experiments conducted on real datasets demonstrate the remarkable superiority of the proposed CDAT in improving the adversarial robustness of the sparse domain. The codes and datasets are available on https://github.com/HymanLoveGIN/CDAT.

  Full text in ACM Digital Library

• RES 🕓15MLoRA: Multi-Domain Low-Rank Adaptive Network for CTR Prediction
  by Zhiming Yang (Northwestern Polytechnical University), Haining Gao (Alibaba Group), Dehong Gao (Northwestern Polytechnical University), Luwei Yang (Alibaba Group), Libin Yang (Northwestern Polytechnical University), Xiaoyan Cai (Northwestern Polytechnical University), Wei Ning (Alibaba Group) and Guannan Zhang (Alibaba Group)

  Click-through rate (CTR) prediction is one of the fundamental tasks in the industry, especially in e-commerce, social media, and streaming media. It directly impacts website revenues, user satisfaction, and user retention. However, real-world production platforms often encompass various domains to cater for diverse customer needs. Traditional CTR prediction models struggle in multi-domain recommendation scenarios, facing challenges of data sparsity and disparate data distributions across domains. Existing multi-domain recommendation approaches introduce specific-domain modules for each domain, which partially address these issues but often significantly increase model parameters and lead to insufficient training. In this paper, we propose a Multi-domain Low-Rank Adaptive network (MLoRA) for CTR prediction, where we introduce a specialized LoRA module for each domain. This approach enhances the model’s performance in multi-domain CTR prediction tasks and is able to be applied to various deep-learning models. We evaluate the proposed method on several multi-domain datasets. Experimental results demonstrate our MLoRA approach achieves a significant improvement compared with state-of-the-art baselines. Furthermore, we deploy it in the production environment of the Alibaba.COM. The online A/B testing results indicate the superiority and flexibility in real-world production environments. The code of our MLoRA is publicly available.

  Full text in ACM Digital Library

• RES 🕓15Instructing and Prompting Large Language Models for Explainable Cross-domain Recommendations
  by Alessandro Petruzzelli (University of Bari Aldo Moro), Cataldo Musto (University of Bari), Lucrezia Laraspata (University of Bari), Ivan Rinaldi (University of Bari Aldo Moro), Marco de Gemmis (University of Bari Aldo Moro), Pasquale Lops (University of Bari) and Giovanni Semeraro (University of Bari)

  In this paper, we present a strategy to provide users with explainable cross-domain recommendations (CDR) that exploits large language models (LLMs). Generally speaking, CDR is a task that is hard to tackle, mainly due to data sparsity issues. Indeed, CDR models require a large amount of data labeled in both source and target domains, which are not easy to collect. Accordingly, our approach relies on the intuition that the knowledge that is already encoded in LLMs can be used to more easily bridge the domains and seamlessly provide users with personalized cross-domain suggestions.

  To this end, we designed a pipeline to: (a) instruct a LLM to handle a CDR task; (b) design a personalized prompt, based on the preferences of the user in a source domain, and a list of items to be ranked in target domain; (c) feed the LLM with the prompt, in both zero-shot and one-shot settings, and process the answer in order to extract the recommendations and a natural language explanation. As shown in the experimental evaluation, our approach beats several established state-of-the-art baselines for CDR in most of the experimental settings, thus showing the effectiveness of LLMs also in this novel and scarcely investigated scenario.

  Full text in ACM Digital Library

• RES 🕓15Cross-Domain Latent Factors Sharing via Implicit Matrix Factorization
  by Abdulaziz Samra (Skolkovo Institute of Science and Technology), Evgeny Frolov (AIRI; Skolkovo Institute of Science and Technology), Alexey Vasilev (Sber), Alexander Grigorevskiy (Independent researcher) and Anton Vakhrushev (Independent researcher)

  Data sparsity has been one of the long-standing problems for recommender systems. One of the solutions to mitigate this issue is to exploit knowledge available in other source domains. However, many cross-domain recommender systems introduce a complex architecture that makes them less scalable in practice. On the other hand, matrix factorization methods are still considered to be strong baselines for single-domain recommendations. In this paper, we introduce the CDIMF, a model that extends the standard implicit matrix factorization with ALS to cross-domain scenarios. We apply the Alternating Direction Method of Multipliers to learn shared latent factors for overlapped users while factorizing the interaction matrix. In a dual-domain setting, experiments on industrial datasets demonstrate a competing performance of CDIMF for both cold-start and warm-start. The proposed model can outperform most other recent cross-domain and single-domain models. We also provide the code to reproduce experiments on GitHub.

  Full text in ACM Digital Library

• RES 🕓15Discerning Canonical User Representation for Cross-Domain Recommendation
  by Siqian Zhao (University at Albany – SUNY) and Sherry Sahebi (University at Albany – SUNY)

  Cross-domain recommender systems (CDRs) aim to enhance recommendation outcomes by information transfer across different domains. Existing CDRs have investigated the learning of both domain-specific and domain-shared user preferences to enhance recommendation performance. However, these models typically allow the disparities between shared and distinct user preferences to emerge freely in any space, lacking sufficient constraints to identify differences between two domains and to ensure that both domains are considered simultaneously. Canonical Correlation Analysis (CCA) has shown promise for transferring information between domains. However, CCA only models domain similarities and fails to capture the potential differences between user preferences in different domains. We propose Discerning Canonical User Representation for Cross-Domain Recommendation (DiCUR-CDR) that learns domain-shared and domain-specific user representations simultaneously considering both domains’ latent spaces. DiCUR-CDR introduces Discerning Canonical Correlation (DisCCA) user representation learning, a novel design of non-linear CCA for mapping user representations. Unlike prior CCA models that only model the domain-shared multivariate representations by finding their linear transformations, DisCCA uses the same transformations to discover the domain-specific representations too. We compare DiCUR-CDR against several state-of-the-art approaches using two real-world datasets and demonstrate the significance of separately learning shared and specific user representations via DisCCA.

  Full text in ACM Digital Library

• IND 🕓10Joint Modeling of Search and Recommendations Via an Unified Contextual Recommender (UniCoRn)
  by Moumita Bhattacharya (Netflix), Vito Ostuni (Netflix) and Sudarshan Lamkhede (Netflix)

  Full text in ACM Digital Library

Back to program