Recent progress in large language model (LLM) based generative recommendation (GR) shows that leveraging LLM world knowledge can substantially improve performance. However, existing methods rely on fixed, manually designed instructions to generate semantic knowledge and directly incorporate it into GR, which has two limitations: (1) fixed instructions cannot capture the multidimensional heterogeneity of user interests; (2) uncontrollable knowledge fusion may conflict with behavioral signals and harm recommendations. To address these limitations, we propose LWGR, a framework that leverages Lagrangian constraints to transfer users' personalized World knowledge from LLMs into Generative Recommendation. LWGR enhances GR along two axes: knowledge extraction and fusion. It builds user personalized soft instructions to extract behavior-relevant LLM world knowledge. Then, it formulates knowledge fusion as an optimization problem with explicitly bounded performance degradation, solved via a Lagrangian primal–dual method that selectively incorporates beneficial knowledge. We further design two training strategies for different LLM scales and a deployment scheme that combines nearline precomputation with lightweight online serving. Experiments on multiple public datasets and one industrial dataset show that LWGR outperforms eight state-of-the-art baselines by up to 11.23% and brings a 1.35% revenue lift on a large-scale advertising platform, demonstrating its effectiveness and practicality.

The storage and recall of factual associations in auto-regressive transformer language models (LMs) have drawn a great deal of attention, inspiring knowledge editing by directly modifying the located model weights. Most editing works achieve knowledge editing under the guidance of existing interpretations of knowledge recall that mainly focus on subject knowledge. However, these interpretations are seriously flawed, neglecting relation information and leading to the *over-generalizing* problem for editing. In this work, we discover a novel relation-focused perspective to interpret the knowledge recall of transformer LMs during inference and apply it on single knowledge editing to avoid over-generalizing. Experimental results on the dataset supplemented with a new R-Specificity criterion demonstrate that our editing approach significantly alleviates over-generalizing while remaining competitive on other criteria, breaking the domination of subject-focused editing for future research.

Graph neural network (GNN) based recommender systems have been widely used in diverse service platforms as they can more effectively capture users' interests. Nevertheless, recent investigations have revealed that the neighborhood aggregation and contrastive learning mechanisms render GNN-based recommender systems more vulnerable to fake profile injection attacks, i.e., shilling attacks. Despite numerous defenses against shilling attacks having emerged, these approaches still face certain challenges, such as the demand for prior knowledge and the difficulty in defending against multiple attacks. Therefore, this paper proposes a two-stage trustworthy GNN-based recommender systems training framework (Trust-GRS), which models the probability of data being fake in a zero-knowledge scenario and establishes a trustworthy neighborhood aggregation and contrastive learning. Through extensive experiments on multiple benchmark datasets against 12 state-of-the-art shilling attacks, we demonstrate that Trust-GRS substantially mitigates the influence of fake data in all attacks, up to 100%, while preserving the original recommendation performance. Benefiting from the absence of the requirement for prior knowledge, Trust-GRS holds significant application value for real-world recommendation platforms.

Parameter-Efficient Tuning (PET) has shown remarkable performance by fine-tuning only a small number of parameters of the pre-trained language models (PLMs) for the downstream tasks, while it is also possible to construct backdoor attacks due to the vulnerability of pretrained weights.However, a large reduction in the number of attackable parameters in PET will cause the user's fine-tuning to greatly affect the effectiveness of backdoor attacks, resulting in backdoor forgetting.We find that the backdoor injection process can be regarded as multitask learning, which has a convergence imbalance problem between the training of clean and poisoned data.And this problem might result in forgetting the backdoor.Based on this finding, we propose a gradient control method to consolidate the attack effect, comprising two strategies.One controls the gradient magnitude distribution cross layers within one task and the other prevents the conflict of gradient directions between tasks.Compared with previous backdoor attack methods in the scenario of PET, our method improves the effect of the attack on sentiment classification and spam detection respectively, which shows that our method is widely applicable to different tasks.

Transformer-based pre-trained language models (PLMs) mostly suffer from excessive overhead despite their advanced capacity.For resource-constrained devices, there is an urgent need for a spatially and temporally efficient model which retains the major capacity of PLMs.However, existing statically compressed models are unaware of the diverse complexities between input instances, potentially resulting in redundancy and inadequacy for simple and complex inputs.Also, miniature models with early exiting encounter challenges in the trade-off between making predictions and serving the deeper layers.Motivated by such considerations, we propose a collaborative optimization for PLMs that integrates static model compression and dynamic inference acceleration.Specifically, the PLM is slenderized in width while the depth remains intact, complementing layer-wise early exiting to speed up inference dynamically.To address the trade-off of early exiting, we propose a joint training approach that calibrates slenderization and preserves contributive structures to each exit instead of only the final layer.Experiments are conducted on GLUE benchmark and the results verify the Pareto optimality of our approach at high compression and acceleration rate with 1/8 parameters and 1/19 FLOPs of BERT.