arXiv Computer Science @arxiv_cs@qoto.org

Current metadata often suffer from incompleteness, inconsistency, and incorrect formatting, hindering effective data reuse and discovery. Using GPT-4 and a metadata knowledge base (CEDAR), we devised a method that standardizes metadata in scientific data sets, ensuring the adherence to community standards. The standardization process involves correcting and refining metadata entries to conform to established guidelines, significantly improving search performance and recall metrics. The investigation uses BioSample and GEO repositories to demonstrate the impact of these enhancements, showcasing how standardized metadata lead to better retrieval outcomes. The average recall improves significantly, rising from 17.65\% with the baseline raw datasets of BioSample and GEO to 62.87\% with our proposed metadata standardization pipeline. This finding highlights the transformative impact of integrating advanced AI models with structured metadata curation tools in achieving more effective and reliable data retrieval.

Modern recommender systems excel at optimizing search result relevance for e-commerce platforms. While maintaining this relevance, platforms seek opportunities to maximize revenue through search result adjustments. To address the trade-off between relevance and revenue, we propose the $\mathsf{RARe}$ ($\textbf{R}$aising $\textbf{A}$dvertisement $\textbf{Re}$venue) framework. $\mathsf{RARe}$ stacks a click model and a reranking model. We train the $\mathsf{RARe}$ framework with a loss function to find revenue and relevance trade-offs. According to our experience, the click model is crucial in the $\mathsf{RARe}$ framework. We propose and compare two different click models that take into account the context of items in a search result. The first click model is a Gradient-Boosting Decision Tree with Concatenation (GBDT-C), which includes a context in the traditional GBDT model for click prediction. The second model, SAINT-Q, adapts the Sequential Attention model to capture influences between search results. Our experiments indicate that the proposed click models outperform baselines and improve the overall quality of our framework. Experiments on the industrial dataset, which will be released publicly, show $\mathsf{RARe}$'s significant revenue improvements while preserving a high relevance.

The essence of modern e-Commercial search system lies in matching user's intent and available candidates depending on user's query, providing personalized and precise service. However, user's query may be incorrect due to ambiguous input and typo, leading to inaccurate search. These cases may be released by query rewrite: modify query to other representation or expansion. However, traditional query rewrite replies on static rewrite vocabulary, which is manually established meanwhile lacks interaction with both domain knowledge in e-Commercial system and common knowledge in the real world. In this paper, with the ability to generate text content of Large Language Models (LLMs), we provide an iterative framework to generate query rewrite. The framework incorporates a 3-stage procedure in each iteration: Rewrite Generation with domain knowledge by Retrieval-Augmented Generation (RAG) and query understanding by Chain-of-Thoughts (CoT); Online Signal Collection with automatic positive rewrite update; Post-training of LLM with multi task objective to generate new rewrites. Our work (named as IterQR) provides a comprehensive framework to generate \textbf{Q}uery \textbf{R}ewrite with both domain / real-world knowledge. It automatically update and self-correct the rewrites during \textbf{iter}ations. \method{} has been deployed in Meituan Delivery's search system (China's leading food delivery platform), providing service for users with significant improvement.

Many e-commerce search pipelines have four stages, namely: retrieval, filtering, ranking, and personalized-reranking. The retrieval stage must be efficient and yield high recall because relevant products missed in the first stage cannot be considered in later stages. This is challenging for task-oriented queries (queries with actionable intent) where user requirements are contextually intensive and difficult to understand. To foster research in the domain of e-commerce, we created a novel benchmark for Task-oriented Queries (TQE) by using LLM, which operates over the existing ESCI product search dataset. Furthermore, we propose a novel method 'Graph-based Recall Improvement for Task-oriented queries' (GRIT) to address the most crucial first-stage recall improvement needs. GRIT leads to robust and statistically significant improvements over state-of-the-art lexical, dense, and learned-sparse baselines. Our system supports both traditional and task-oriented e-commerce queries, yielding up to 6.3% recall improvement. In the indexing stage, GRIT first builds a product-product similarity graph using user clicks or manual annotation data. During retrieval, it locates neighbors with higher contextual and action relevance and prioritizes them over the less relevant candidates from the initial retrieval. This leads to a more comprehensive and relevant first-stage result set that improves overall system recall. Overall, GRIT leverages the locality relationships and contextual insights provided by the graph using neighboring nodes to enrich the first-stage retrieval results. We show that the method is not only robust across all introduced parameters, but also works effectively on top of a variety of first-stage retrieval methods.

This article introduces the Data Retrieval Web Engine (also referred to as doctor web), a flexible and modular tool for extracting structured data from web pages using a simple query language. We discuss the engineering challenges addressed during its development, such as dynamic content handling and messy data extraction. Furthermore, we cover the steps for making the DR Web Engine public, highlighting its open source potential.

Retrieval-Augmented Generation (RAG), by integrating non-parametric knowledge from external knowledge bases into models, has emerged as a promising approach to enhancing response accuracy while mitigating factual errors and hallucinations. This method has been widely applied in tasks such as Question Answering (QA). However, existing RAG methods struggle with open-domain QA tasks because they perform independent retrieval operations and directly incorporate the retrieved information into generation without maintaining a summarizing memory or using adaptive retrieval strategies, leading to noise from redundant information and insufficient information integration. To address these challenges, we propose Adaptive memory-based optimization for enhanced RAG (Amber) for open-domain QA tasks, which comprises an Agent-based Memory Updater, an Adaptive Information Collector, and a Multi-granular Content Filter, working together within an iterative memory updating paradigm. Specifically, Amber integrates and optimizes the language model's memory through a multi-agent collaborative approach, ensuring comprehensive knowledge integration from previous retrieval steps. It dynamically adjusts retrieval queries and decides when to stop retrieval based on the accumulated knowledge, enhancing retrieval efficiency and effectiveness. Additionally, it reduces noise by filtering irrelevant content at multiple levels, retaining essential information to improve overall model performance. We conduct extensive experiments on several open-domain QA datasets, and the results demonstrate the superiority and effectiveness of our method and its components. The source code is available \footnote{https://anonymous.4open.science/r/Amber-B203/}.

Sequential recommendation is an advanced recommendation technique that utilizes the sequence of user behaviors to generate personalized suggestions by modeling the temporal dependencies and patterns in user preferences. However, it requires a server to centrally collect users' data, which poses a threat to the data privacy of different users. In recent years, federated learning has emerged as a distributed architecture that allows participants to train a global model while keeping their private data locally. This survey pioneers Federated Sequential Recommendation (FedSR), where each user joins as a participant in federated training to achieve a recommendation service that balances data privacy and model performance. We begin with an introduction to the background and unique challenges of FedSR. Then, we review existing solutions from two levels, each of which includes two specific techniques. Additionally, we discuss the critical challenges and future research directions in FedSR.

Generative recommendation has emerged as a promising paradigm aiming at directly generating the identifiers of the target candidates. Most existing methods attempt to leverage prior knowledge embedded in Pre-trained Language Models (PLMs) to improve the recommendation performance. However, they often fail to accommodate the differences between the general linguistic knowledge of PLMs and the specific needs of recommendation systems. Moreover, they rarely consider the complementary knowledge between the multimodal information of items, which represents the multi-faceted preferences of users. To facilitate efficient recommendation knowledge transfer, we propose a novel approach called Multimodal Quantitative Language for Generative Recommendation (MQL4GRec). Our key idea is to transform items from different domains and modalities into a unified language, which can serve as a bridge for transferring recommendation knowledge. Specifically, we first introduce quantitative translators to convert the text and image content of items from various domains into a new and concise language, known as quantitative language, with all items sharing the same vocabulary. Then, we design a series of quantitative language generation tasks to enrich quantitative language with semantic information and prior knowledge. Finally, we achieve the transfer of recommendation knowledge from different domains and modalities to the recommendation task through pre-training and fine-tuning. We evaluate the effectiveness of MQL4GRec through extensive experiments and comparisons with existing methods, achieving improvements over the baseline by 11.18\%, 14.82\%, and 7.95\% on the NDCG metric across three different datasets, respectively.

Explainable recommender systems are designed to elucidate the explanation behind each recommendation, enabling users to comprehend the underlying logic. Previous works perform rating prediction and explanation generation in a multi-task manner. However, these works suffer from incoherence between predicted ratings and explanations. To address the issue, we propose a novel framework that employs a large language model (LLM) to generate a rating, transforms it into a rating vector, and finally generates an explanation based on the rating vector and user-item information. Moreover, we propose utilizing publicly available LLMs and pre-trained sentiment analysis models to automatically evaluate the coherence without human annotations. Extensive experimental results on three datasets of explainable recommendation show that the proposed framework is effective, outperforming state-of-the-art baselines with improvements of 7.3\% in explainability and 4.4\% in text quality.

Composed Image Retrieval (CIR) aims to search an image of interest using a combination of a reference image and modification text as the query. Despite recent advancements, this task remains challenging due to limited training data and laborious triplet annotation processes. To address this issue, this paper proposes to synthesize the training triplets to augment the training resource for the CIR problem. Specifically, we commence by training a modification text generator exploiting large-scale multimodal models and scale up the CIR learning throughout both the pretraining and fine-tuning stages. During pretraining, we leverage the trained generator to directly create Modification Text-oriented Synthetic Triplets(MTST) conditioned on pairs of images. For fine-tuning, we first synthesize reverse modification text to connect the target image back to the reference image. Subsequently, we devise a two-hop alignment strategy to incrementally close the semantic gap between the multimodal pair and the target image. We initially learn an implicit prototype utilizing both the original triplet and its reversed version in a cycle manner, followed by combining the implicit prototype feature with the modification text to facilitate accurate alignment with the target image. Extensive experiments validate the efficacy of the generated triplets and confirm that our proposed methodology attains competitive recall on both the CIRR and FashionIQ benchmarks.

Parallelization has become a cornerstone of modern computing, influencing everything from high performance supercomputers to everyday mobile devices. This paper presents a comprehensive guide on the fundamentals of parallelization that every computer scientist should know, beginning with a historical perspective that traces the evolution from early theoretical models such as PRAM and BSP to today's advanced multicore and heterogeneous architectures. We explore essential theoretical frameworks, practical paradigms, and synchronization mechanisms while discussing implementation strategies using processes, threads, and modern models like the Actor framework. Additionally, we examine how hardware components including CPUs, caches, memory, and accelerators interact with software to impact performance, scalability, and load balancing. This work demystifies parallel programming by integrating historical context, theoretical underpinnings, and practical case studies. It equips readers with the tools to design, optimize, and troubleshoot parallel applications in an increasingly concurrent computing landscape.

We introduce AIBrix, a cloud-native, open-source framework designed to optimize and simplify large-scale LLM deployment in cloud environments. Unlike traditional cloud-native stacks, AIBrix follows a co-design philosophy, ensuring every layer of the infrastructure is purpose-built for seamless integration with inference engines like vLLM. AIBrix introduces several key innovations to reduce inference costs and enhance performance including high-density LoRA management for dynamic adapter scheduling, LLM-specific autoscalers, and prefix-aware, load-aware routing. To further improve efficiency, AIBrix incorporates a distributed KV cache, boosting token reuse across nodes, leading to a 50% increase in throughput and a 70% reduction in inference latency. AIBrix also supports unified AI runtime which streamlines model management while maintaining vendor-agnostic engine compatibility. For large-scale multi-node inference, AIBrix employs hybrid orchestration -- leveraging Kubernetes for coarse-grained scheduling and Ray for fine-grained execution -- to balance efficiency and flexibility. Additionally, an SLO-driven GPU optimizer dynamically adjusts resource allocations, optimizing heterogeneous serving to maximize cost efficiency while maintaining service guarantees. Finally, AIBrix enhances system reliability with AI accelerator diagnostic tools, enabling automated failure detection and mock-up testing to improve fault resilience. AIBrix is available at https://github.com/vllm-project/aibrix.

The French company EDF uses supervisory control and data acquisition systems in conjunction with a data management platform to monitor hydropower plant, allowing engineers and technicians to analyse the time-series collected. Depending on the strategic importance of the monitored hydropower plant, the number of time-series collected can vary greatly making it difficult to generate valuable information from the extracted data. In an attempt to provide an answer to this particular problem, a condition detection and diagnosis method combining clustering algorithms and autoencoder neural networks for pattern recognition has been developed and is presented in this paper. First, a dimension reduction algorithm is used to create a 2-or 3-dimensional projection that allows the users to identify unsuspected relationships between datapoints. Then, a collection of clustering algorithms regroups the datapoints into clusters. For each identified cluster, an autoencoder neural network is trained on the corresponding dataset. The aim is to measure the reconstruction error between each autoencoder model and the measured values, thus creating a proximity index for each state discovered during the clustering stage.

BoxRL-NNV is a Python tool for the detection of safety violations in neural networks by computing the bounds of the output variables, given the bounds of the input variables of the network. This is done using global extrema estimation via Latin Hypercube Sampling, and further refinement using L-BFGS-B for local optimization around the initial guess. This paper presents an overview of BoxRL-NNV, as well as our results for a subset of the ACAS Xu benchmark. A complete evaluation of the tool's performance, including benchmark comparisons with state-of-the-art tools, shall be presented at the Sixth International Verification of Neural Networks Competition (VNN-COMP'25).

Large language models have been widely deployed in various applications, encompassing both interactive online tasks and batched offline tasks. Given the burstiness and latency sensitivity of online tasks, over-provisioning resources is common practice. This allows for the integration of latency-insensitive offline tasks during periods of low online load, enhancing resource utilization. However, strategically serving online and offline tasks through a preemption mechanism fails to fully leverage the flexibility of offline tasks and suffers from KV cache recomputation and irregular workloads. In this paper, we introduce Echo, a collaborative online-offline task serving system, including a scheduler, a KV cache manager, and estimation toolkits. The scheduler and KV cache manager work tightly to maximize the throughput of offline tasks, while the estimator further predicts execution time to ensure online task SLOs. The scheduler leverages the batch information of last iteration to reduce the search space for finding the optimal schedule. The KV cache manager sets the priority of the KV cache based on the type of tasks and the opportunity of prefix sharing to reduce the recomputation. Finally, the estimation toolkits predict the execution time, future memory consumption, and the throughput of offline tasks to guide the scheduler, KV cache manager, and the system deployer. Evaluation based on real-world workloads demonstrates that Echo can increase offline task throughput by up to $3.3\times$, while satisfying online task SLOs.

Air traffic analytics systems are pivotal for ensuring safety, efficiency, and predictability in air travel. However, traditional systems struggle to handle the increasing volume and complexity of air traffic data. This project explores the application of real-time big data processing frameworks like Apache Spark, HDFS, and Spark Streaming for developing a new robust system. By reviewing existing research on real-time systems and analyzing the challenges and opportunities presented by big data technologies, we propose an architecture for a real-time system. Our project pipeline involves real-time data collection from flight information sources through flight API's, ingestion into Kafka, and transmission to Elasticsearch for visualization using Kibana. Additionally, we present a dashboard of U.S. airlines on PowerBI, demonstrating the potential of real-time analytics in revolutionizing air traffic management.

The emergence of new, off-path smart network cards (SmartNICs), known generally as Data Processing Units (DPU), has opened a wide range of research opportunities. Of particular interest is the use of these and related devices in tandem with their host's CPU, creating a heterogeneous computing system with new properties and strengths to be explored, capable of accelerating a wide variety of workloads. This survey begins by providing background information to this new field, such as discussing its origins, its motivations and challenges, listing a few of the current market offerings for DPUs, and providing some brief information about the major programming languages and frameworks for using them. Then, we review and categorize a number of recent works in the field, covering a wide variety of studies, benchmarks, and application areas such as in data center infrastructure, commercial uses, and AI and ML acceleration.

Running deep learning models on resource-constrained edge devices has drawn significant attention due to its fast response, privacy preservation, and robust operation regardless of Internet connectivity. While these devices already cope with various intelligent tasks, the latest edge devices that are equipped with multiple types of low-power accelerators (i.e., both mobile GPU and NPU) can bring another opportunity; a task that used to be too heavy for an edge device in the single-accelerator world might become viable in the upcoming heterogeneous-accelerator world.To realize the potential in the context of 3D object detection, we identify several technical challenges and propose PointSplit, a novel 3D object detection framework for multi-accelerator edge devices that addresses the problems. Specifically, our PointSplit design includes (1) 2D semantics-aware biased point sampling, (2) parallelized 3D feature extraction, and (3) role-based group-wise quantization. We implement PointSplit on TensorFlow Lite and evaluate it on a customized hardware platform comprising both mobile GPU and EdgeTPU. Experimental results on representative RGB-D datasets, SUN RGB-D and Scannet V2, demonstrate that PointSplit on a multi-accelerator device is 24.7 times faster with similar accuracy compared to the full-precision, 2D-3D fusion-based 3D detector on a GPU-only device.

Transformer models have revolutionized a wide spectrum of disciplines, especially in language processing. The recent success has proven that model size scalability is crucial for achieving superior performance metrics. However, training large transformer models is challenging even on modern hardware with powerful GPUs and high-speed interconnects. Existing studies primarily focus on optimizing model training distribution strategies to minimize memory footprint and enhance training speed, often overlooking the scalability challenges related to model size and hardware constraints. To address this oversight, we thoroughly investigate computational, memory, and network demands of training large transformers using the Fully Sharded Data Parallel (FSDP) distributed strategy across different hardware clusters. We explore the intricate relationships between model size and hardware setups to identify configurations that ensure maximum model and hardware efficiency, effective sequence length management, and optimal training throughput. A significant finding of our study is the critical interplay of the cluster's connection bandwidth and GPU memory size compared to the computational performance of GPUs. This interplay limits training efficiency, underscoring the role of both hardware characteristics as a possible bottleneck. By integrating theoretical analysis with simulations and empirical tests, we demonstrate how hardware limitations affect training efficacy, identifying key hardware thresholds and the impact of network connectivity. Our findings prompt a reassessment of training strategies guiding users on the way to finding hardware-optimal FSDP configurations, enhancing training efficiency for large-scale transformer models.

Edge computing environments increasingly rely on lightweight container orchestration platforms to manage resource-constrained devices. This paper provides an empirical analysis of five lightweight kubernetes distributions (KD)(k0s, k3s, KubeEdge, OpenYurt, and Kubernetes (k8s)) focusing on their performance and resource efficiency in edge computing scenarios. We evaluated key metrics such as CPU, memory, disk usage, throughput, and latency under varying workloads, utilizing a testbed of Intel NUCs and Raspberry Pi devices. Our results demonstrate significant differences in performance: k3s exhibited the lowest resource consumption, while k0s and k8s excelled in data plane throughput and latency. Under heavy stress scenarios, k3s and k0s accomplished the same workloads faster than the other distributions. OpenYurt offered balanced performance, suitable for hybrid cloud-edge use cases, but was less efficient in terms of resource usage and scalability compared to k0s, k3s and k8s. KubeEdge, although feature-rich for edge environments, exhibited higher resource consumption and lower scalability. These findings offer valuable insights for developers and operators selecting appropriate KD based on specific performance and resource efficiency requirements for edge computing environments.