Big Data最新文献

Automated program repair (APR) has been studied extensively in recent years. Existing approaches mainly generate single-position patches that fail to address multilocation faults effectively. While existing multistep repair approaches can iteratively generate patches for each fault position sequentially, their data augmentation methodologies lack rationality and deviate from real-world scenarios. Furthermore, they overlook the interdependencies between faulty statements, leading to patches learned from erroneous contextual patterns. In this article, we propose MuTemAPR, an APR approach that iteratively generates multilocation patches. MuTemAPR incorporates templates with neural machine translation. Specifically, our method introduces three key innovations. First, we design a template-based data augmentation framework that transforms single-line faulty code into multilocation faulty code through 35 mutation templates. It simulates a real-world environment by establishing variable-type mapping tables for more accurate repair augmentation. Second, we propose a reinforced faulty context training method that employs progressive annotation to incrementally learn repair processes from top to bottom in multifault code. Third, we implement a semantic constraint mechanism during training that enforces syntactic and semantic rules through differential analysis between templates, input code, and generated patches. We evaluate MuTemAPR on the widely used Defects4j benchmark. Experimental results demonstrate that our approach can effectively repair multilocation faults, successfully fixing five additional bugs compared with state-of-the-art methods on Defects4j v1.2 and v2.0.

Engineering teams need timely signals about evolving requirements and release risk, yet multilingual fan discourse around live sports is noisy, code-switched, and saturated with sarcasm and event-driven drift. We present Hybrid DeepSentX, an AI-driven framework that converts crowd commentary into actionable requirements insight and sprint-level risk scores. The pipeline couples multilingual transformer encoders with an inductive GraphSAGE conversation graph to inject relational context across posts, and adds a reinforcement learner whose reward is shaped to prioritize correct decisions on sarcasm-heavy items and rapidly shifting events. We assembled a million-plus posts from X, Reddit, and sports forums and evaluated the framework against strong baselines, including BERT, long short-term memory, support-vector machines, and recent hybrid models, with significance tests, calibration analysis, ablations, and efficiency profiling. DeepSentX achieved higher macro-averaged accuracy and F1 on code-switched and sarcastic subsets, reduced missed risk flags, and produced developer-facing artefacts that directly support backlog grooming and defect triage. Relative to prior hybrids that combine transformers with either graph reasoning or reinforcement alone, our contributions are fourfold: (i) a unified multilingual design that integrates transformer, graph, and reinforcement components for sarcasm and drift robustness, (ii) an annotated multi-platform corpus with explicit code switching and sarcasm labels and per platform language balance, (iii) a rigorous comparative study reporting accuracy, calibration, latency, memory, and parameter count, and (iv) deployment artefacts that turn model outputs into requirement clusters and sprint risk scores suitable for continuous planning.

We introduce VARM, variant relationship matcher strategy, to identify pairs of variant products in e-commerce catalogs. Traditional definitions of entity resolution are concerned with whether product mentions refer to the same underlying product. However, this fails to capture product relationships that are critical for e-commerce applications, such as having similar, but not identical, products listed on the same webpage or share reviews. Here, we formulate a new type of entity resolution in variant product relationships to capture these similar e-commerce product links. In contrast with the traditional definition, the new definition requires both identifying if two products are variant matches of each other and what the attributes are that vary between them. To satisfy these two requirements, we developed a strategy that leverages the strengths of both encoding and generative AI models. First, we construct a dataset that captures webpage product links, and therefore variant product relationships, to train an encoding large language model (LLM) to predict variant matches for any given pair of products. Second, we use retrieval-augmented generation-prompted generative LLMs to extract variation and common attributes amongst groups of variant products. To validate our strategy, we evaluated model performance using real data from one of the world's leading e-commerce retailers. The results showed that our strategy outperforms alternative solutions and paves the way to exploiting these new types of product relationships.

Dysarthria, a motor speech disorder characterized by slurred and often unintelligible speech, presents substantial challenges for effective communication. Conventional automatic speech recognition systems frequently underperform on dysarthric speech, particularly in severe cases. To address this gap, we introduce low-latency acoustic transcription and textual encoding (LATTE), an advanced framework designed for real-time dysarthric speech recognition. LATTE integrates preprocessing, acoustic processing, and transcription mapping into a unified pipeline, with its core powered by a hybrid architecture that combines convolutional layers for acoustic feature extraction with bidirectional temporal layers for modeling temporal dependencies. Evaluated on the UA-Speech dataset, LATTE achieves a word error rate of 12.5%, phoneme error rate of 8.3%, and a character error rate of 1%. By enabling accurate, low-latency transcription of impaired speech, LATTE provides a robust foundation for enhancing communication and accessibility in both digital applications and real-time interactive environments.

Named entity recognition (NER) is a core task in natural language processing that identifies and classifies entities, such as people, organizations, and locations within text. It has traditionally been applied in areas like text summarization, machine translation, and question answering. In recent years, NER has gained growing importance in health care, where electronic clinical records and online platforms generate large amounts of unstructured medical data. However, applying NER in clinical contexts introduces unique challenges due to the complexity of medical terminology and the need for high accuracy. In this study, we focused on the development of a real-time, low-latency NER system designed for cross-lingual speech-to-text applications, with a particular emphasis on cancer therapy-related clinical records and traditional Chinese medicine (TCM). We explored the integration of deep learning (DL) architectures optimized for low-latency neural processing to extract structured information from multilingual spoken content in medical settings, particularly in multimodal environments. We evaluate DL-based methods and propose a semi-supervised approach that combines TCM-specific corpora with biomedical resources to improve recognition accuracy. The findings provide both a systematic review of current methods and practical insights for building real-time clinical applications that support decision-making and information management in health care.

This study intends to identify the critical factors that shape college students' adoption of AI-generated news, with a specific focus on integrating Big Data methodologies into the Technology Acceptance Model (TAM) framework. Building on TAM, the research incorporates "trust" as a core variable to develop a dual-path theoretical model that combines technological cognition (e.g., perceived usefulness, perceived ease of use) and psychological emotions. Unlike traditional TAM-based studies relying solely on questionnaire data, this research enriches its data sources by leveraging Big Data techniques-including the collection and analysis of college students' real-time behavioral data (e.g., AI news reading duration, sharing frequency, source verification clicks) and unstructured text data (e.g., sentiment orientation in comment sections)-to complement the survey data from 300 college students. Through a questionnaire survey of 300 college students and data analysis using the structural equation model, the study found that trust has the strongest direct positive impact on the willingness to use (β = 0.49, p < 0.001), and its influence is significantly greater than perceived usefulness (β = 0.35, p < 0.001). Meanwhile, although perceived ease of use does not directly affect the willingness to use, it has significant indirect effects by enhancing trust and perceived usefulness. The results show that in the AI news context with high-risk perception, trust is a more crucial psychological mechanism than traditional technological cognitive factors. These findings have expanded the explanatory boundaries of the TAM model in new technology fields and provided empirical evidence and practical inspiration for AI developers to optimize system credibility and for educators to conduct algorithmic literacy training.

Optical coherence tomography (OCT) offers significant advantages of noncontact operation, high resolution, and real-time imaging, making it particularly suitable for acquiring human retinal images and playing a crucial role in diagnosing and monitoring retinal diseases such as diabetic macular edema (DME). OCT is a key noninvasive imaging modality for retinal diseases such as DME, offering high-resolution visualization of retinal layers and fluid accumulations. However, retinal fluid segmentation faces several challenges including variations in fluid size, location, and shape, as well as complex irregular boundaries. To address these issues, we propose TL-TransUNet, a novel lightweight segmentation model based on TransUNet. The model incorporates a hybrid self-attention mechanism that effectively combines linear self-attention with residual filtered multilayer perceptron modules, reducing both parameter size and computational complexity while capturing global relationships and local details to improve segmentation performance for small lesions. Furthermore, the decoder employs wavelet convolution that utilizes wavelet transform to extract multi-scale features from low- to high-frequency components, enhancing the model's multi-scale learning capability. Experimental results on a public DME dataset demonstrate that our proposed method outperforms several mainstream segmentation approaches, demonstrating superior performance.

This study presents a real-time, context-adaptive advertisement (ad in short) recommendation framework that dynamically updates user context and utilizes a multistage ranking and filtering pipeline to deliver highly relevant and personalized ads. Contextual ads contribute to better conversion rates and play a significant role in e-commerce. In contrast, non-contextual ads engender frustration among advertisers and users: commercialization efforts frequently prove ineffective due to poor user engagement, as evidenced by high ad-skipping rates. The current practices in digital advertising involve non-contextual and irrelevant ads, which result in poor conversion rates. To address this problem, this article explores semantically enriched and context-aware recommender systems, aiming to align ads with user interests. The proposed framework investigates various components, including a user context extractor (UCE), recommender system, ads database, ads ranker, and ads filter. This study also explores how high-quality and relevant content, along with clickable advertising, contributes to improving customer relationships and reducing ad avoidance. During contextual augmentation, ads that become relevant and engaging are projected to have increased click-through rates in a real-world application. Customer engagement and satisfaction would also increase due to a reduction in ad fatigue and the delivery of relevant content. Furthermore, it can curb ad avoidance because users will gladly respond to ads that suit their interests. Businesses make higher conversions because the more relevant recommendation means greater user interaction. The proposed framework combines a UCE, an ad database, a ranking mechanism, and a filtering module to deliver real-time, personalized recommendations. Evaluated using a k-nearest neighbor-based model, the system achieved improved precision (from 0.8275 to 0.9283), recall (from 0.4628 to 0.5201), and normalized discounted cumulative gain (from 0.9906 to 0.9915). These gains demonstrate that integrating fine-grained, dynamic user context substantially enhances recommendation quality and user engagement, offering a scalable foundation for intelligent, adaptive advertising systems. This research contributes toward the future development of an AI-enabled advertising strategy, with an emphasis on dynamic ad targeting that goes hand in hand with personalization and thus improved conversion rate.