Recent years have witnessed the prosperity of Knowledge Graph-Based Recommendation System (KGRS), which enriches the representation of users, items, and entities by structural knowledge with striking improvement. Nevertheless, its unaffordable computational cost still limits researchers from exploring more sophisticated models. We observe that the bottleneck for training efficiency arises from the knowledge graph, which is plagued by the well-known issue of knowledge explosion. Recently, some works have attempted to slim the inflated KG via summarization techniques, which summarize multiple real nodes into the single virtual one. However, these summarized virtual nodes may ignore collaborative signals and thus fail to figure out the redundant nodes related to recommendation task. To this end, in this article, we propose a novel approach called KGTrimmer for knowledge graph pruning tailored for recommendation, to remove the unessential nodes while minimizing performance degradation. Specifically, we design an importance evaluator from a dual-view perspective. For the collective view, we embrace the idea of collective intelligence by extracting community consensus based on abundant collaborative signals, i.e., nodes are considered important if they attract attention of numerous users. For the holistic view, we learn a global mask to identify the valueless nodes from their inherent properties or overall popularity. With the collective and holistic importance scores, we build an end-to-end importance-aware graph neural network, which injects filtered knowledge to enhance the distillation of valuable user-item collaborative signals. Ultimately, we generate a pruned knowledge graph with lightweight, stable, and robust properties to facilitate the following-up recommendation task. Extensive experiments are conducted on three publicly available datasets to prove the effectiveness and generalizability of KGTrimmer, where it can reduce the number of triplets in KG by up to 90% without compromising performance.

With the rapid development of Large Language Models (LLMs), aligning these models with human preferences and values is critical to ensuring ethical and safe applications. However, existing alignment techniques such as RLHF or DPO often require direct fine-tuning on LLMs with billions of parameters, resulting in substantial computational costs and inefficiencies. To address this, we propose Micro token-level Accept-Reject Aligning (MARA) approach designed to operate independently of the language models. MARA simplifies the alignment process by decomposing sentence-level preference learning into token-level binary classification, where a compact three-layer fully-connected network determines whether candidate tokens are “Accepted” or “Rejected” as part of the response. Extensive experiments across seven different LLMs and three open-source datasets show that MARA achieves significant improvements in alignment performance while reducing computational costs. The source code and implementation details are publicly available at https://github.com/IAAR-Shanghai/MARA, and the trained models are released at https://huggingface.co/IAAR-Shanghai/MARA_AGENTS.

Commentary provides readers with a deep understanding of events by presenting diverse arguments and evidence. However, creating commentary is a time-consuming task, even for skilled commentators. Large language models (LLMs) have simplified the process of natural language generation, but their direct application in commentary creation still faces challenges due to unique task requirements. These requirements can be categorized into two levels: 1) fundamental requirements, which include creating well-structured and logically consistent narratives, and 2) advanced requirements, which involve generating quality arguments and providing convincing evidence. In this paper, we introduce Xinyu, an efficient LLM-based system designed to assist commentators in generating Chinese commentaries. To meet the fundamental requirements, we deconstruct the generation process into sequential steps, proposing targeted strategies and supervised fine-tuning (SFT) for each step. To address the advanced requirements, we present an argument ranking model for arguments and establish a comprehensive evidence database that includes up-to-date events and classic books, thereby strengthening the substantiation of the evidence with retrieval augmented generation (RAG) technology. To evaluate the generated commentaries more fairly, corresponding to the two-level requirements, we introduce a comprehensive evaluation metric that considers five distinct perspectives in commentary generation. Our experiments confirm the effectiveness of our proposed system. We also observe a significant increase in the efficiency of commentators in real-world scenarios, with the average time spent on creating a commentary dropping from 4 hours to 20 minutes. Importantly, such an increase in efficiency does not compromise the quality of the commentaries.

Dynamic share recommendation, which aims at recommending a friend who would like to share a particular item at a certain timestamp, has emerged as a novel task for social-oriented e-commerce platforms. Different from traditional graph-based recommendation tasks, with integrating the interconnected social interactions and fine-grained temporal information from historical share records, this novel task may encounter one unique challenge, i.e., how to deal with the dynamic social connections and asymmetric share interactions. Even worse, users may keep inactive during some periods, which results in difficulties in updating personalized profiles. To address the above challenges, in this paper, we propose a dynamic graph share recommendation model called DynShare. Specifically, we first divide each user embedding into two parts, namely the invitation embedding and vote embedding to show the tendencies of sending and receiving items, respectively. Then, temporal graph attention networks (TGATs) based on bi-directional continuous time dynamic graphs (CTDGs) are leveraged to encode temporal neighbor information from different directions. Afterward, to estimate how different users perceive the time intervals after the last interaction, we further design a time-interval aware personalized projection operator on the foundation of temporal point processes (TPPs) to project user embedding for the next-time share prediction. Extensive experiments on a real-world e-commerce share dataset have demonstrated that our proposed DynShare can achieve better results compared with state-of-the-art baseline methods. And our code is available on the project website: https://github.com/meteor-gif/DynShare.

This paper studies an unmanned aerial vehicle (UAV)-assisted wireless powered IoT network, where a rotary-wing UAV adopts fly-hover-communicate protocol to successively visit IoT devices in demand. During the hovering periods, the UAV works on full-duplex mode to simultaneously collect data from the target device and charge other devices within its coverage. Practical propulsion power consumption model and non-linear energy harvesting model are taken into account. We formulate a multi-objective optimization problem to jointly optimize three objectives: maximization of sum data rate, maximization of total harvested energy and minimization of UAV's energy consumption over a particular mission period. These three objectives are in conflict with each other partly and weight parameters are given to describe associated importance. Since IoT devices keep gathering information from the physical surrounding environment and their requirements to upload data change dynamically, online path planning of the UAV is required. In this paper, we apply deep reinforcement learning algorithm to achieve online decision. An extended deep deterministic policy gradient (DDPG) algorithm is proposed to learn control policies of UAV over multiple objectives. While training, the agent learns to produce optimal policies under given weights conditions on the basis of achieving timely data collection according to the requirement priority and avoiding devices' data overflow. The verification results show that the proposed MODDPG (multi-objective DDPG) algorithm achieves joint optimization of three objectives and optimal policies can be adjusted according to weight parameters among optimization objectives.

Transmission rate and harvested energy are well-known conflictive optimization objectives in simultaneous wireless information and power transfer (SWIPT) systems, and thus their trade-off and joint optimization are important problems to be studied. In this paper, we investigate joint power allocation and splitting control in a SWIPT-enabled non-orthogonal multiple access (NOMA) system with the power splitting (PS) technique, with an aim to optimize the total transmission rate and harvested energy simultaneously whilst satisfying the minimum rate and the harvested energy requirements of each user. These two conflicting objectives make the formulated problem a constrained multi-objective optimization problem. Since the harvested power is usually stored in the battery and used to support the reverse link transmission, we transform the harvested energy into throughput and define a new objective function by summing the weighted values of the transmission rate achieved by information decoding and transformed throughput from energy harvesting, defined as equivalent-sum-rate (ESR). As a result, the original problem is transformed into a single-objective optimization problem. The considered ESR maximization problem which involves joint optimization of power allocation and PS ratio is nonconvex, and hence challenging to solve. In order to tackle it, we decouple the original nonconvex problem into two convex subproblems and solve them iteratively. In addition, both equal PS ratio case and independent PS ratio case are considered to further explore the performance. Numerical results validate the theoretical findings and demonstrate that significant performance gain over the traditional rate maximization scheme can be achieved by the proposed algorithms in a SWIPT-enabled NOMA system.