Recommendation systems are key to boosting user engagement,satisfaction,and retention,particularly on media platforms where personalized content is vital.Sequential recommendation systems learn from user-item interact...Recommendation systems are key to boosting user engagement,satisfaction,and retention,particularly on media platforms where personalized content is vital.Sequential recommendation systems learn from user-item interactions to predict future items of interest.However,many current methods rely on unique user and item IDs,limiting their ability to represent users and items effectively,especially in zero-shot learning scenarios where training data is scarce.With the rapid development of Large Language Models(LLMs),researchers are exploring their potential to enhance recommendation systems.However,there is a semantic gap between the linguistic semantics of LLMs and the collaborative semantics of recommendation systems,where items are typically indexed by IDs.Moreover,most research focuses on item representations,neglecting personalized user modeling.To address these issues,we propose a sequential recommendation framework using LLMs,called CIT-Rec,a model that integrates Collaborative semantics for user representation and Image and Text information for item representation to enhance Recommendations.Specifically,by aligning intuitive image information with text containing semantic features,we can more accurately represent items,improving item representation quality.We focus not only on item representations but also on user representations.To more precisely capture users’personalized preferences,we use traditional sequential recommendation models to train on users’historical interaction data,effectively capturing behavioral patterns.Finally,by combining LLMs and traditional sequential recommendation models,we allow the LLM to understand linguistic semantics while capturing collaborative semantics.Extensive evaluations on real-world datasets show that our model outperforms baseline methods,effectively combining user interaction history with item visual and textual modalities to provide personalized recommendations.展开更多
Model evaluation using benchmark datasets is an important method to measure the capability of large language models(LLMs)in specific domains,and it is mainly used to assess the knowledge and reasoning abilities of LLM...Model evaluation using benchmark datasets is an important method to measure the capability of large language models(LLMs)in specific domains,and it is mainly used to assess the knowledge and reasoning abilities of LLMs.Therefore,in order to better assess the capability of LLMs in the agricultural domain,Agri-Eval was proposed as a benchmark for assessing the knowledge and reasoning ability of LLMs in agriculture.The assessment dataset used in Agri-Eval covered seven major disciplines in the agricultural domain:crop science,horticulture,plant protection,animal husbandry,forest science,aquaculture science,and grass science,and contained a total of 2283 questions.Among domestic general-purpose LLMs,DeepSeek R1 performed best with an accuracy rate of 75.49%.In the realm of international general-purpose LLMs,Gemini 2.0 pro exp 0205 standed out as the top performer,achieving an accuracy rate of 74.28%.As an LLMs in agriculture vertical,Shennong V2.0 outperformed all the LLMs in China,and the answer accuracy rate of agricultural knowledge exceeded that of all the existing general-purpose LLMs.The launch of Agri-Eval helped the LLM developers to comprehensively evaluate the model's capability in the field of agriculture through a variety of tasks and tests to promote the development of the LLMs in the field of agriculture.展开更多
Classifying job offers into occupational categories is a fundamental task in human resource information systems,as it improves and streamlines indexing,search,and matching between openings and job seekers.Comprehensiv...Classifying job offers into occupational categories is a fundamental task in human resource information systems,as it improves and streamlines indexing,search,and matching between openings and job seekers.Comprehensive occupational databases such as O∗NET or ESCO provide detailed taxonomies of interrelated positions that can be leveraged to align the textual content of postings with occupational categories,thereby facilitating standardization,cross-system interoperability,and access to metadata for each occupation(e.g.,tasks,knowledge,skills,and abilities).In this work,we explore the effectiveness of fine-tuning existing language models(LMs)to classify job offers with occupational descriptors from O∗NET.This enables a more precise assessment of candidate suitability by identifying the specific knowledge and skills required for each position,and helps automate recruitment processes by mitigating human bias and subjectivity in candidate selection.We evaluate three representative BERT-like models:BERT,RoBERTa,and DeBERTa.BERT serves as the baseline encoder-only architecture;RoBERTa incorporates advances in pretraining objectives and data scale;and DeBERTa introduces architectural improvements through disentangled attention mechanisms.The best performance was achieved with the DeBERTa model,although the other models also produced strong results,and no statistically significant differences were observed acrossmodels.We also find that these models typically reach optimal performance after only a few training epochs,and that training with smaller,balanced datasets is effective.Consequently,comparable results can be obtained with models that require fewer computational resources and less training time,facilitating deployment and practical use.展开更多
Background:Assess ChatGPT and Bard's effectiveness in the initial identification of articles for Otolaryngology—Head and Neck Surgery systematic literature reviews.Methods:Three PRISMA-based systematic reviews(Ja...Background:Assess ChatGPT and Bard's effectiveness in the initial identification of articles for Otolaryngology—Head and Neck Surgery systematic literature reviews.Methods:Three PRISMA-based systematic reviews(Jabbour et al.2017,Wong et al.2018,and Wu et al.2021)were replicated using ChatGPTv3.5 and Bard.Outputs(author,title,publication year,and journal)were compared to the original references and cross-referenced with medical databases for authenticity and recall.Results:Several themes emerged when comparing Bard and ChatGPT across the three reviews.Bard generated more outputs and had greater recall in Wong et al.'s review,with a broader date range in Jabbour et al.'s review.In Wu et al.'s review,ChatGPT-2 had higher recall and identified more authentic outputs than Bard-2.Conclusion:Large language models(LLMs)failed to fully replicate peer-reviewed methodologies,producing outputs with inaccuracies but identifying relevant,especially recent,articles missed by the references.While human-led PRISMA-based reviews remain the gold standard,refining LLMs for literature reviews shows potential.展开更多
This study demonstrates a novel integration of large language models,machine learning,and multicriteria decision-making to investigate self-moderation in small online communities,a topic under-explored compared to use...This study demonstrates a novel integration of large language models,machine learning,and multicriteria decision-making to investigate self-moderation in small online communities,a topic under-explored compared to user behavior and platform-driven moderation on social media.The proposed methodological framework(1)utilizes large language models for social media post analysis and categorization,(2)employs k-means clustering for content characterization,and(3)incorporates the TODIM(Tomada de Decisão Interativa Multicritério)method to determine moderation strategies based on expert judgments.In general,the fully integrated framework leverages the strengths of these intelligent systems in a more systematic evaluation of large-scale decision problems.When applied in social media moderation,this approach promotes nuanced and context-sensitive self-moderation by taking into account factors such as cultural background and geographic location.The application of this framework is demonstrated within Facebook groups.Eight distinct content clusters encompassing safety,harassment,diversity,and misinformation are identified.Analysis revealed a preference for content removal across all clusters,suggesting a cautious approach towards potentially harmful content.However,the framework also highlights the use of other moderation actions,like account suspension,depending on the content category.These findings contribute to the growing body of research on self-moderation and offer valuable insights for creating safer and more inclusive online spaces within smaller communities.展开更多
The malicious dissemination of hate speech via compromised accounts,automated bot networks and malware-driven social media campaigns has become a growing cybersecurity concern.Automatically detecting such content in S...The malicious dissemination of hate speech via compromised accounts,automated bot networks and malware-driven social media campaigns has become a growing cybersecurity concern.Automatically detecting such content in Spanish is challenging due to linguistic complexity and the scarcity of annotated resources.In this paper,we compare two predominant AI-based approaches for the forensic detection of malicious hate speech:(1)finetuning encoder-only models that have been trained in Spanish and(2)In-Context Learning techniques(Zero-and Few-Shot Learning)with large-scale language models.Our approach goes beyond binary classification,proposing a comprehensive,multidimensional evaluation that labels each text by:(1)type of speech,(2)recipient,(3)level of intensity(ordinal)and(4)targeted group(multi-label).Performance is evaluated using an annotated Spanish corpus,standard metrics such as precision,recall and F1-score and stability-oriented metrics to evaluate the stability of the transition from zero-shot to few-shot prompting(Zero-to-Few Shot Retention and Zero-to-Few Shot Gain)are applied.The results indicate that fine-tuned encoder-only models(notably MarIA and BETO variants)consistently deliver the strongest and most reliable performance:in our experiments their macro F1-scores lie roughly in the range of approximately 46%–66%depending on the task.Zero-shot approaches are much less stable and typically yield substantially lower performance(observed F1-scores range approximately 0%–39%),often producing invalid outputs in practice.Few-shot prompting(e.g.,Qwen 38B,Mistral 7B)generally improves stability and recall relative to pure zero-shot,bringing F1-scores into a moderate range of approximately 20%–51%but still falling short of fully fine-tuned models.These findings highlight the importance of supervised adaptation and discuss the potential of both paradigms as components in AI-powered cybersecurity and malware forensics systems designed to identify and mitigate coordinated online hate campaigns.展开更多
Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Re...Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Recent approaches such as Distilling Step-by-Step(DSbS)introduce explanation supervision,yet they apply it in a uniform manner that may not fully exploit the different learning dynamics of prediction and explanation.In this work,we propose a task-structured curriculum learning(TSCL)framework that structures training into three sequential phases:(i)prediction-only,to establish stable feature representations;(ii)joint prediction-explanation,to align task outputs with rationale generation;and(iii)explanation-only,to refine the quality of rationales.This design provides a simple but effective modification to DSbS,requiring no architectural changes and adding negligible training cost.We justify the phase scheduling with ablation studies and convergence analysis,showing that an initial prediction-heavy stage followed by a balanced joint phase improves both stability and explanation alignment.Extensive experiments on five datasets(e-SNLI,ANLI,CommonsenseQA,SVAMP,and MedNLI)demonstrate that TSCL consistently outperforms strong baselines,achieving gains of+1.7-2.6 points in accuracy and 0.8-1.2 in ROUGE-L,corresponding to relative error reductions of up to 21%.Beyond lexical metrics,human evaluation and ERASERstyle faithfulness diagnostics confirm that TSCL produces more faithful and informative explanations.Comparative training curves further reveal faster convergence and lower variance across seeds.Efficiency analysis shows less than 3%overhead in wall-clock training time and no additional inference cost,making the approach practical for realworld deployment.This study demonstrates that a simple task-structured curriculum can significantly improve the effectiveness of knowledge distillation.By separating and sequencing objectives,TSCL achieves a better balance between accuracy,stability,and explanation quality.The framework generalizes across domains,including medical NLI,and offers a principled recipe for future applications in multimodal reasoning and reinforcement learning.展开更多
The collection and annotation of lar ge-scale bird datasets are resource-intensive and time-consuming processes that significantly limit the scalability and accuracy of biodiversity monitoring systems.While self-super...The collection and annotation of lar ge-scale bird datasets are resource-intensive and time-consuming processes that significantly limit the scalability and accuracy of biodiversity monitoring systems.While self-supervised learning(SSL)has emerged as a promising approach for leveraging unannotated data,current SSL methods face two critical challenges in bird species recognition:(1)long-tailed data distributions that result in poor performance on underrepresented species;and(2)domain shift issues caused by data augmentation strategies designed to mitigate class imbalance.Here we present SDNet,a novel SSL-based bird recognition framework that integrates diffusion models with large language models(LLMs)to overcome these limitations.SDNet employs LLMs to generate semantically rich textual descriptions for tail-class species by prompting the models with species taxonomy,morphological attributes,and habitat information,producing detailed natural language priors that capture fine-grained visual characteristics(e.g.,plumage patterns,body proportions,and distinctive markings).These textual descriptions are subsequently used by a conditional diffusion model to synthesize new bird image samples through cross-attention mechanisms that fuse textual embeddings with intermediate visual feature representations during the denoising process,ensuring generated images preserve species-specific morphological details while maintaining photorealistic quality.Additionally,we incorporate a Swin Transformer as the feature extraction backbone whose hierarchical window-based attention mechanism and shifted windowing scheme enable multi-scale local feature extraction that proves particularly effective at capturing finegrained discriminative patterns(such as beak shape and feather texture)while mitigating domain shift between synthetic and original images through consistent feature representations across both data sources.SDNet is validated on both a self-constructed dataset(Bird_BXS)an d a publicly available benchmark(Birds_25),demonstrating substantial improvements over conventional SSL approaches.Our results indicate that the synergistic integration of LLMs,diffusion models,and the Swin Transformer architecture contributes significantly to recognition accuracy,particularly for rare and morphologically similar species.These findings highlight the potential of SDNet for addressing fundamental limitations of existing SSL methods in avian recognition tasks and establishing a new paradigm for efficient self-supervised learning in large-scale ornithological vision applications.展开更多
LargeLanguageModels(LLMs)are increasingly appliedinthe fieldof code translation.However,existing evaluation methodologies suffer from two major limitations:(1)the high overlap between test data and pretraining corpora...LargeLanguageModels(LLMs)are increasingly appliedinthe fieldof code translation.However,existing evaluation methodologies suffer from two major limitations:(1)the high overlap between test data and pretraining corpora,which introduces significant bias in performance evaluation;and(2)mainstream metrics focus primarily on surface-level accuracy,failing to uncover the underlying factors that constrain model capabilities.To address these issues,this paper presents TCode(Translation-Oriented Code Evaluation benchmark)—a complexity-controllable,contamination-free benchmark dataset for code translation—alongside a dedicated static feature sensitivity evaluation framework.The dataset is carefully designed to control complexity along multiple dimensions—including syntactic nesting and expression intricacy—enabling both broad coverage and fine-grained differentiation of sample difficulty.This design supports precise evaluation of model capabilities across a wide spectrum of translation challenges.The proposed evaluation framework introduces a correlation-driven analysis mechanism based on static program features,enabling predictive modeling of translation success from two perspectives:Code Form Complexity(e.g.,code length and character density)and Semantic Modeling Complexity(e.g.,syntactic depth,control-flow nesting,and type system complexity).Empirical evaluations across representative LLMs—including Qwen2.5-72B and Llama3.3-70B—demonstrate that even state-of-the-art models achieve over 80% compilation success on simple samples,but their accuracy drops sharply below 40% on complex cases.Further correlation analysis indicates that Semantic Modeling Complexity alone is correlated with up to 60% of the variance in translation success,with static program features exhibiting nonlinear threshold effects that highlight clear capability boundaries.This study departs fromthe traditional accuracy-centric evaluation paradigm and,for the first time,systematically characterizes the capabilities of large languagemodels in translation tasks through the lens of programstatic features.The findings provide actionable insights for model refinement and training strategy development.展开更多
War rehearsals have become increasingly important in national security due to the growing complexity of international affairs.However,traditional rehearsal methods,such as military chess simulations,are inefficient an...War rehearsals have become increasingly important in national security due to the growing complexity of international affairs.However,traditional rehearsal methods,such as military chess simulations,are inefficient and inflexible,with particularly pronounced limitations in command and decision-making.The overwhelming volume of information and high decision complexity hinder the realization of autonomous and agile command and control.To address this challenge,an intelligent warfare simulation framework named Command-Agent is proposed,which deeply integrates large language models(LLMs)with digital twin battlefields.By constructing a highly realistic battlefield environment through real-time simulation and multi-source data fusion,the natural language interaction capabilities of LLMs are leveraged to lower the command threshold and to enable autonomous command through the Observe-Orient-Decide-Act(OODA)feedback loop.Within the Command-Agent framework,a multimodel collaborative architecture is further adopted to decouple the decision-generation and command-execution functions of LLMs.By combining specialized models such as Deep Seek-R1 and MCTool,the limitations of single-model capabilities are overcome.MCTool is a lightweight execution model fine-tuned for military Function Calling tasks.The framework also introduces a Vector Knowledge Base to mitigate hallucinations commonly exhibited by LLMs.Experimental results demonstrate that Command-Agent not only enables natural language-driven simulation and control but also deeply understands commander intent.Leveraging the multi-model collaborative architecture,during red-blue UAV confrontations involving 2 to 8 UAVs,the integrated score is improved by an average of 41.8%compared to the single-agent system(MCTool),accompanied by a 161.8%optimization in the battle loss ratio.Furthermore,when compared with multi-agent systems lacking the knowledge base,the inclusion of the Vector Knowledge Base further improves overall performance by 16.8%.In comparison with the general model(Qwen2.5-7B),the fine-tuned MCTool leads by 5%in execution efficiency.Therefore,the proposed Command-Agent introduces a novel perspective to the military command system and offers a feasible solution for intelligent battlefield decision-making.展开更多
Objective To develop a clinical decision and prescription generation system(CDPGS)specifically for diarrhea in traditional Chinese medicine(TCM),utilizing a specialized large language model(LLM),Qwen-TCM-Dia,to standa...Objective To develop a clinical decision and prescription generation system(CDPGS)specifically for diarrhea in traditional Chinese medicine(TCM),utilizing a specialized large language model(LLM),Qwen-TCM-Dia,to standardize diagnostic processes and prescription generation.Methods Two primary datasets were constructed:an evaluation benchmark and a fine-tuning dataset consisting of fundamental diarrhea knowledge,medical records,and chain-ofthought(CoT)reasoning datasets.After an initial evaluation of 16 open-source LLMs across inference time,accuracy,and output quality,Qwen2.5 was selected as the base model due to its superior overall performance.We then employed a two-stage low-rank adaptation(LoRA)fine-tuning strategy,integrating continued pre-training on domain-specific knowledge with instruction fine-tuning using CoT-enriched medical records.This approach was designed to embed the clinical logic(symptoms→pathogenesis→therapeutic principles→prescriptions)into the model’s reasoning capabilities.The resulting fine-tuned model,specialized for TCM diarrhea,was designated as Qwen-TCM-Dia.Model performance was evaluated for disease diagnosis and syndrome type differentiation using accuracy,precision,recall,and F1-score.Furthermore,the quality of the generated prescriptions was compared with that of established open-source TCM LLMs.Results Qwen-TCM-Dia achieved peak performance compared to both the base Qwen2.5 model and five other open-source TCM LLMs.It achieved 97.05%accuracy and 91.48%F1-score in disease diagnosis,and 74.54%accuracy and 74.21%F1-score in syndrome type differentiation.Compared with existing open-source TCM LLMs(BianCang,HuangDi,LingDan,TCMLLM-PR,and ZhongJing),Qwen-TCM-Dia exhibited higher fidelity in reconstructing the“symptoms→pathogenesis→therapeutic principles→prescriptions”logic chain.It provided complete prescriptions,whereas other models often omitted dosages or generated mismatched prescriptions.Conclusion By integrating continued pre-training,CoT reasoning,and a two-stage fine-tuning strategy,this study establishes a CDPGS for diarrhea in TCM.The results demonstrate the synergistic effect of strengthening domain representation through pre-training and activating logical reasoning via CoT.This research not only provides critical technical support for the standardized diagnosis and treatment of diarrhea but also offers a scalable paradigm for the digital inheritance of expert TCM experience and the intelligent transformation of TCM.展开更多
Large language models(LLMs)have revolutionized AI applications across diverse domains.However,their widespread deployment has introduced critical security vulnerabilities,particularly prompt injection attacks that man...Large language models(LLMs)have revolutionized AI applications across diverse domains.However,their widespread deployment has introduced critical security vulnerabilities,particularly prompt injection attacks that manipulate model behavior through malicious instructions.Following Kitchenham’s guidelines,this systematic review synthesizes 128 peer-reviewed studies from 2022 to 2025 to provide a unified understanding of this rapidly evolving threat landscape.Our findings reveal a swift progression from simple direct injections to sophisticated multimodal attacks,achieving over 90%success rates against unprotected systems.In response,defense mechanisms show varying effectiveness:input preprocessing achieves 60%–80%detection rates and advanced architectural defenses demonstrate up to 95%protection against known patterns,though significant gaps persist against novel attack vectors.We identified 37 distinct defense approaches across three categories,but standardized evaluation frameworks remain limited.Our analysis attributes these vulnerabilities to fundamental LLM architectural limitations,such as the inability to distinguish instructions from data and attention mechanism vulnerabilities.This highlights critical research directions such as formal verification methods,standardized evaluation protocols,and architectural innovations for inherently secure LLM designs.展开更多
Objective To develop QingNangTCM,a specialized large language model(LLM)tailored for expert-level traditional Chinese medicine(TCM)question-answering and clinical reasoning,addressing the scarcity of domain-specific c...Objective To develop QingNangTCM,a specialized large language model(LLM)tailored for expert-level traditional Chinese medicine(TCM)question-answering and clinical reasoning,addressing the scarcity of domain-specific corpora and specialized alignment.Methods We constructed QnTCM_Dataset,a corpus of 100000 entries,by integrating data from ShenNong_TCM_Dataset and SymMap v2.0,and synthesizing additional samples via retrieval-augmented generation(RAG)and persona-driven generation.The dataset comprehensively covers diagnostic inquiries,prescriptions,and herbal knowledge.Utilizing P-Tuning v2,we fine-tuned the GLM-4-9B-Chat backbone to develop QingNangTCM.A multidimensional evaluation framework,assessing accuracy,coverage,consistency,safety,professionalism,and fluency,was established using metrics such as bilingual evaluation understudy(BLEU),recall-oriented understudy for gisting evaluation(ROUGE),metric for evaluation of translation with explicit ordering(METEOR),and LLM-as-a-Judge with expert review.Qualitative analysis was conducted across four simulated clinical scenarios:symptom analysis,disease treatment,herb inquiry,and failure cases.Baseline models included GLM-4-9BChat,DeepSeek-V2,HuatuoGPT-II(7B),and GLM-4-9B-Chat(freeze-tuning).Results QingNangTCM achieved the highest scores in BLEU-1/2/3/4(0.425/0.298/0.137/0.064),ROUGE-1/2(0.368/0.157),and METEOR(0.218),demonstrating a balanced and superior normalized performance profile of 0.900 across the dimensions of accuracy,coverage,and consistency.Although its ROUGE-L score(0.299)was lower than that of HuatuoGPT-II(7B)(0.351),it significantly outperformed domain-specific models in expert-validated win rates for professionalism(86%)and safety(73%).Qualitative analysis confirmed that the model strictly adheres to the“symptom-syndrome-pathogenesis-treatment”reasoning chain,though occasional misclassifications and hallucinations persisted when dealing with rare medicinal materials and uncommon syndromes.展开更多
Online Public Opinion Reports consolidate news and social media for timely crisis management by governments and enterprises.While large language models(LLMs)enable automated report generation,this specific domain lack...Online Public Opinion Reports consolidate news and social media for timely crisis management by governments and enterprises.While large language models(LLMs)enable automated report generation,this specific domain lacks formal task definitions and corresponding benchmarks.To bridge this gap,we define the Automated Online Public Opinion Report Generation(OPOR-Gen)task and construct OPOR-Bench,an event-centric dataset with 463 crisis events across 108 countries(comprising 8.8 K news articles and 185 K tweets).To evaluate report quality,we propose OPOR-Eval,a novel agent-based framework that simulates human expert evaluation.Validation experiments show OPOR-Eval achieves a high Spearman’s correlation(ρ=0.70)with human judgments,though challenges in temporal reasoning persist.This work establishes an initial foundation for advancing automated public opinion reporting research.展开更多
Background:Large language models(LLMs)have shown considerable promise in supporting clinical decision-making.However,their adoption and evaluation in dermatology remains limited.This study aimed to explore the prefere...Background:Large language models(LLMs)have shown considerable promise in supporting clinical decision-making.However,their adoption and evaluation in dermatology remains limited.This study aimed to explore the preferences of Chinese dermatologists regarding LLM-generated responses in clinical psoriasis scenarios and to assess how they prioritize key quality dimensions,including accuracy,traceability,and logicality.Methods:A cross-sectional,web-based survey was conducted between December 25,2024,and January 22,2025,following the Checklist for Reporting Results of Internet E-Surveys guidelines.A total of 1247 valid responses were collected from practicing dermatologists across 33 of China's provincial-level administrative divisions.Participants evaluated responses to five categories of clinical questions(etiology,clinical presentation,differential diagnosis,treatment,and case study)generated by five LLMs:ChatGPT-4o,Kimi.ai,Doubao,ZuoYiGPT,and Lingyi-agent.Statistical associations between participant characteristics and model preferences were examined using chi-square tests.Results:ChatGPT-4o(Model 1)emerged as the most preferred model across all clinical tasks,consistently receiving the highest number of votes in case study(n=740),clinical presentation(n=666),differential diagnosis(n=707),etiology(n=602),and treatment(n=656).Significant variation in model preference by professional title was observed only for the differential diagnosis task(χ^(2)=21.13,df=12,p=0.0485),while no significant differences were found across hospital tiers(p>0.05).In terms of evaluation dimensions,accuracy was most frequently rated as“very important”(n=635).A significant association existed between hospital tier and the most valued dimension(χ^(2)=27.667,df=9,p=0.0011),with dermatologists in primary hospitals prioritizing traceability more than their peers in higher-tier hospitals.No significant associations were found across professional titles(p=0.127).Conclusions:Chinese dermatologists suggest a strong preference for ChatGPT-4o over domestic LLMs in psoriasis-related clinical tasks.While accuracy remains the primary criterion,traceability and logicality are also critical,particularly for clinicians in lower-tier hospitals.These findings suggest that future clinical LLMs should prioritize not only content accuracy but also source transparency and structural clarity to meet the diverse needs of different clinical settings.展开更多
Large language models(LLMs)show considerable potential to revolutionize healthcare through their performance across diverse clinical applications.Given the inherent constraints of LLMs and the critical nature of medic...Large language models(LLMs)show considerable potential to revolutionize healthcare through their performance across diverse clinical applications.Given the inherent constraints of LLMs and the critical nature of medical practice,a rigorous and systematic evaluation of their medical competence is imperative.This study presents a comprehensive review of the established methodologies and benchmarks for evaluating the medical competence of LLMs,encompassing a thorough analysis of current assessment practices across medical knowledge,clinical practice competence,and ethical-safety considerations.By integrating clinician competency assessment frameworks into LLMs evaluation,we propose a structured tri-dimensional framework that systematically organizes existing evaluation approaches according to medical theoretical knowledge,clinical practice ability,and ethical-safety considerations.Furthermore,this research provides critical insights into future developmental trajectories while establishing foundational frameworks and standardization protocols for the integration of LLMs into medical practice.展开更多
Conversational recommender systems(CRSs)focus on refining preferences and providing personalized recommendations through natural language interactions and dialogue history.Large language models(LLMs)have shown outstan...Conversational recommender systems(CRSs)focus on refining preferences and providing personalized recommendations through natural language interactions and dialogue history.Large language models(LLMs)have shown outstanding performance across various domains,thereby prompting researchers to investigate their applicability in recommendation systems.However,due to the lack of task-specific knowledge and an inefficient feature extraction process,LLMs still have suboptimal performance in recommendation tasks.Therefore,external knowledge sources,such as knowledge graphs(KGs)and knowledge bases(KBs),are often introduced to address the issue of data sparsity.Compared to KGs,KBs possess higher retrieval efficiency,making them more suitable for scenarios where LLMs serve as recommenders.To this end,we introduce a novel framework integrating LLMs with KBs for enhanced retrieval generation,namely LLMKB.LLMKB initially leverages structured knowledge to create mapping dictionaries,extracting entity-relation information from heterogeneous knowledge to construct KBs.Then,LLMKB achieves the embedding calibration between user information representations and documents in KBs through retrieval model fine-tuning.Finally,LLMKB employs retrievalaugmented generation to produce recommendations based on fused text inputs,followed by post-processing.Experiment results on two public CRS datasets demonstrate the effectiveness of our framework.Our code is publicly available at the link:https://anonymous.4open.science/r/LLMKB-6FD0.展开更多
Fundamental physics often confronts complex symbolic problems with few guiding exemplars or established principles.While artificial intelligence(AI)offers promise,its typical need for vast datasets to learn from hinde...Fundamental physics often confronts complex symbolic problems with few guiding exemplars or established principles.While artificial intelligence(AI)offers promise,its typical need for vast datasets to learn from hinders its use in these information-scarce frontiers.We introduce learning at criticality(LaC),a reinforcement learning scheme that tunes large language models(LLMs)to a sharp learning transition,addressing this information scarcity.At this transition,LLMs achieve peak generalization from minimal data,exemplified by 7-digit base-7 addition-a test of nontrivial arithmetic reasoning.To elucidate this peak,we analyze a minimal concept-network model designed to capture the essence of how LLMs might link tokens.Trained on a single exemplar,this model also undergoes a sharp learning transition.This transition exhibits hallmarks of a second-order phase transition,notably power-law distributed solution path lengths.At this critical point,the system maximizes a“critical thinking pattern”crucial for generalization,enabled by the underlying scale-free exploration.This suggests LLMs reach peak performance by operating at criticality,where such explorative dynamics enable the extraction of underlying operational rules.We demonstrate LaC in quantum field theory:an 8B-parameter LLM,tuned to its critical point by LaC using a few exemplars of symbolic Matsubara sums,solves unseen,higher-order problems,significantly outperforming far larger models.LaC thus leverages critical phenomena,a physical principle,to empower AI for complex,data-sparse challenges in fundamental physics.展开更多
We analyze the suitability of existing pre-trained transformer-based language models(PLMs)for abstractive text summarization on German technical healthcare texts.The study focuses on the multilingual capabilities of t...We analyze the suitability of existing pre-trained transformer-based language models(PLMs)for abstractive text summarization on German technical healthcare texts.The study focuses on the multilingual capabilities of these models and their ability to perform the task of abstractive text summarization in the healthcare field.The research hypothesis was that large language models could perform high-quality abstractive text summarization on German technical healthcare texts,even if the model is not specifically trained in that language.Through experiments,the research questions explore the performance of transformer language models in dealing with complex syntax constructs,the difference in performance between models trained in English and German,and the impact of translating the source text to English before conducting the summarization.We conducted an evaluation of four PLMs(GPT-3,a translation-based approach also utilizing GPT-3,a German language Model,and a domain-specific bio-medical model approach).The evaluation considered the informativeness using 3 types of metrics based on Recall-Oriented Understudy for Gisting Evaluation(ROUGE)and the quality of results which is manually evaluated considering 5 aspects.The results show that text summarization models could be used in the German healthcare domain and that domain-independent language models achieved the best results.The study proves that text summarization models can simplify the search for pre-existing German knowledge in various domains.展开更多
BACKGROUND Gastrointestinal diseases have complex etiologies and clinical presentations.An accurate diagnosis requires physicians to integrate diverse information,including medical history,laboratory test results,and ...BACKGROUND Gastrointestinal diseases have complex etiologies and clinical presentations.An accurate diagnosis requires physicians to integrate diverse information,including medical history,laboratory test results,and imaging findings.Existing artificial intelligence-assisted diagnostic tools are limited to single-modality information,resulting in recommendations that are often incomplete and may be associated with clinical or legal risks.AIM To develop and evaluate a collaborative multimodal large language model(LLM)framework for clinical decision-making in digestive diseases.METHODS In this observational study,DeepGut,a multimodal LLM collaborative diagnostic framework,was developed to integrate four distinct large models into a four-tiered structure.The framework sequentially accomplishes multimodal infor-mation extraction,logical“chain”construction,diagnostic and treatment suggestion generation,and risk analysis.The model was evaluated using objective metrics,which assess the reliability and comprehensiveness of model-generated results,and subjective expert opinions,which examine the effectiveness of the framework in assisting physicians.RESULTS The diagnostic and treatment recommendations generated by the DeepGut framework achieved exceptional performance,with a diagnostic accuracy of 97.8%,diagnostic completeness of 93.9%,treatment plan accuracy of 95.2%,and treatment plan completeness of 98.0%,significantly surpassing the capabilities of single-modal LLM-based diagnostic tools.Experts evaluating the framework commended the completeness,relevance,and logical coherence of its outputs.However,the collaborative multimodal LLM approach resulted in increased input and output token counts,leading to higher computational costs and extended diagnostic times.CONCLUSION The framework achieves successful integration of multimodal diagnostic data,demonstrating enhanced performance enabled by multimodal LLM collaboration,which opens new horizons for the clinical application of artificial intelligence-assisted technology.展开更多
基金supported by the National Key R&D Program of China[2022YFF0902703]the State Administration for Market Regulation Science and Technology Plan Project(2024MK033).
文摘Recommendation systems are key to boosting user engagement,satisfaction,and retention,particularly on media platforms where personalized content is vital.Sequential recommendation systems learn from user-item interactions to predict future items of interest.However,many current methods rely on unique user and item IDs,limiting their ability to represent users and items effectively,especially in zero-shot learning scenarios where training data is scarce.With the rapid development of Large Language Models(LLMs),researchers are exploring their potential to enhance recommendation systems.However,there is a semantic gap between the linguistic semantics of LLMs and the collaborative semantics of recommendation systems,where items are typically indexed by IDs.Moreover,most research focuses on item representations,neglecting personalized user modeling.To address these issues,we propose a sequential recommendation framework using LLMs,called CIT-Rec,a model that integrates Collaborative semantics for user representation and Image and Text information for item representation to enhance Recommendations.Specifically,by aligning intuitive image information with text containing semantic features,we can more accurately represent items,improving item representation quality.We focus not only on item representations but also on user representations.To more precisely capture users’personalized preferences,we use traditional sequential recommendation models to train on users’historical interaction data,effectively capturing behavioral patterns.Finally,by combining LLMs and traditional sequential recommendation models,we allow the LLM to understand linguistic semantics while capturing collaborative semantics.Extensive evaluations on real-world datasets show that our model outperforms baseline methods,effectively combining user interaction history with item visual and textual modalities to provide personalized recommendations.
文摘Model evaluation using benchmark datasets is an important method to measure the capability of large language models(LLMs)in specific domains,and it is mainly used to assess the knowledge and reasoning abilities of LLMs.Therefore,in order to better assess the capability of LLMs in the agricultural domain,Agri-Eval was proposed as a benchmark for assessing the knowledge and reasoning ability of LLMs in agriculture.The assessment dataset used in Agri-Eval covered seven major disciplines in the agricultural domain:crop science,horticulture,plant protection,animal husbandry,forest science,aquaculture science,and grass science,and contained a total of 2283 questions.Among domestic general-purpose LLMs,DeepSeek R1 performed best with an accuracy rate of 75.49%.In the realm of international general-purpose LLMs,Gemini 2.0 pro exp 0205 standed out as the top performer,achieving an accuracy rate of 74.28%.As an LLMs in agriculture vertical,Shennong V2.0 outperformed all the LLMs in China,and the answer accuracy rate of agricultural knowledge exceeded that of all the existing general-purpose LLMs.The launch of Agri-Eval helped the LLM developers to comprehensively evaluate the model's capability in the field of agriculture through a variety of tasks and tests to promote the development of the LLMs in the field of agriculture.
文摘Classifying job offers into occupational categories is a fundamental task in human resource information systems,as it improves and streamlines indexing,search,and matching between openings and job seekers.Comprehensive occupational databases such as O∗NET or ESCO provide detailed taxonomies of interrelated positions that can be leveraged to align the textual content of postings with occupational categories,thereby facilitating standardization,cross-system interoperability,and access to metadata for each occupation(e.g.,tasks,knowledge,skills,and abilities).In this work,we explore the effectiveness of fine-tuning existing language models(LMs)to classify job offers with occupational descriptors from O∗NET.This enables a more precise assessment of candidate suitability by identifying the specific knowledge and skills required for each position,and helps automate recruitment processes by mitigating human bias and subjectivity in candidate selection.We evaluate three representative BERT-like models:BERT,RoBERTa,and DeBERTa.BERT serves as the baseline encoder-only architecture;RoBERTa incorporates advances in pretraining objectives and data scale;and DeBERTa introduces architectural improvements through disentangled attention mechanisms.The best performance was achieved with the DeBERTa model,although the other models also produced strong results,and no statistically significant differences were observed acrossmodels.We also find that these models typically reach optimal performance after only a few training epochs,and that training with smaller,balanced datasets is effective.Consequently,comparable results can be obtained with models that require fewer computational resources and less training time,facilitating deployment and practical use.
文摘Background:Assess ChatGPT and Bard's effectiveness in the initial identification of articles for Otolaryngology—Head and Neck Surgery systematic literature reviews.Methods:Three PRISMA-based systematic reviews(Jabbour et al.2017,Wong et al.2018,and Wu et al.2021)were replicated using ChatGPTv3.5 and Bard.Outputs(author,title,publication year,and journal)were compared to the original references and cross-referenced with medical databases for authenticity and recall.Results:Several themes emerged when comparing Bard and ChatGPT across the three reviews.Bard generated more outputs and had greater recall in Wong et al.'s review,with a broader date range in Jabbour et al.'s review.In Wu et al.'s review,ChatGPT-2 had higher recall and identified more authentic outputs than Bard-2.Conclusion:Large language models(LLMs)failed to fully replicate peer-reviewed methodologies,producing outputs with inaccuracies but identifying relevant,especially recent,articles missed by the references.While human-led PRISMA-based reviews remain the gold standard,refining LLMs for literature reviews shows potential.
基金funded by the Office of the Vice-President for Research and Development of Cebu Technological University.
文摘This study demonstrates a novel integration of large language models,machine learning,and multicriteria decision-making to investigate self-moderation in small online communities,a topic under-explored compared to user behavior and platform-driven moderation on social media.The proposed methodological framework(1)utilizes large language models for social media post analysis and categorization,(2)employs k-means clustering for content characterization,and(3)incorporates the TODIM(Tomada de Decisão Interativa Multicritério)method to determine moderation strategies based on expert judgments.In general,the fully integrated framework leverages the strengths of these intelligent systems in a more systematic evaluation of large-scale decision problems.When applied in social media moderation,this approach promotes nuanced and context-sensitive self-moderation by taking into account factors such as cultural background and geographic location.The application of this framework is demonstrated within Facebook groups.Eight distinct content clusters encompassing safety,harassment,diversity,and misinformation are identified.Analysis revealed a preference for content removal across all clusters,suggesting a cautious approach towards potentially harmful content.However,the framework also highlights the use of other moderation actions,like account suspension,depending on the content category.These findings contribute to the growing body of research on self-moderation and offer valuable insights for creating safer and more inclusive online spaces within smaller communities.
基金the research project LaTe4PoliticES(PID2022-138099OB-I00)funded by MCIN/AEI/10.13039/501100011033 and the European Fund for Regional Development(ERDF)-a way to make Europe.Tomás Bernal-Beltrán is supported by University of Murcia through the predoctoral programme.
文摘The malicious dissemination of hate speech via compromised accounts,automated bot networks and malware-driven social media campaigns has become a growing cybersecurity concern.Automatically detecting such content in Spanish is challenging due to linguistic complexity and the scarcity of annotated resources.In this paper,we compare two predominant AI-based approaches for the forensic detection of malicious hate speech:(1)finetuning encoder-only models that have been trained in Spanish and(2)In-Context Learning techniques(Zero-and Few-Shot Learning)with large-scale language models.Our approach goes beyond binary classification,proposing a comprehensive,multidimensional evaluation that labels each text by:(1)type of speech,(2)recipient,(3)level of intensity(ordinal)and(4)targeted group(multi-label).Performance is evaluated using an annotated Spanish corpus,standard metrics such as precision,recall and F1-score and stability-oriented metrics to evaluate the stability of the transition from zero-shot to few-shot prompting(Zero-to-Few Shot Retention and Zero-to-Few Shot Gain)are applied.The results indicate that fine-tuned encoder-only models(notably MarIA and BETO variants)consistently deliver the strongest and most reliable performance:in our experiments their macro F1-scores lie roughly in the range of approximately 46%–66%depending on the task.Zero-shot approaches are much less stable and typically yield substantially lower performance(observed F1-scores range approximately 0%–39%),often producing invalid outputs in practice.Few-shot prompting(e.g.,Qwen 38B,Mistral 7B)generally improves stability and recall relative to pure zero-shot,bringing F1-scores into a moderate range of approximately 20%–51%but still falling short of fully fine-tuned models.These findings highlight the importance of supervised adaptation and discuss the potential of both paradigms as components in AI-powered cybersecurity and malware forensics systems designed to identify and mitigate coordinated online hate campaigns.
文摘Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Recent approaches such as Distilling Step-by-Step(DSbS)introduce explanation supervision,yet they apply it in a uniform manner that may not fully exploit the different learning dynamics of prediction and explanation.In this work,we propose a task-structured curriculum learning(TSCL)framework that structures training into three sequential phases:(i)prediction-only,to establish stable feature representations;(ii)joint prediction-explanation,to align task outputs with rationale generation;and(iii)explanation-only,to refine the quality of rationales.This design provides a simple but effective modification to DSbS,requiring no architectural changes and adding negligible training cost.We justify the phase scheduling with ablation studies and convergence analysis,showing that an initial prediction-heavy stage followed by a balanced joint phase improves both stability and explanation alignment.Extensive experiments on five datasets(e-SNLI,ANLI,CommonsenseQA,SVAMP,and MedNLI)demonstrate that TSCL consistently outperforms strong baselines,achieving gains of+1.7-2.6 points in accuracy and 0.8-1.2 in ROUGE-L,corresponding to relative error reductions of up to 21%.Beyond lexical metrics,human evaluation and ERASERstyle faithfulness diagnostics confirm that TSCL produces more faithful and informative explanations.Comparative training curves further reveal faster convergence and lower variance across seeds.Efficiency analysis shows less than 3%overhead in wall-clock training time and no additional inference cost,making the approach practical for realworld deployment.This study demonstrates that a simple task-structured curriculum can significantly improve the effectiveness of knowledge distillation.By separating and sequencing objectives,TSCL achieves a better balance between accuracy,stability,and explanation quality.The framework generalizes across domains,including medical NLI,and offers a principled recipe for future applications in multimodal reasoning and reinforcement learning.
基金supported by the National Natural Science Foundation of China(32471964)。
文摘The collection and annotation of lar ge-scale bird datasets are resource-intensive and time-consuming processes that significantly limit the scalability and accuracy of biodiversity monitoring systems.While self-supervised learning(SSL)has emerged as a promising approach for leveraging unannotated data,current SSL methods face two critical challenges in bird species recognition:(1)long-tailed data distributions that result in poor performance on underrepresented species;and(2)domain shift issues caused by data augmentation strategies designed to mitigate class imbalance.Here we present SDNet,a novel SSL-based bird recognition framework that integrates diffusion models with large language models(LLMs)to overcome these limitations.SDNet employs LLMs to generate semantically rich textual descriptions for tail-class species by prompting the models with species taxonomy,morphological attributes,and habitat information,producing detailed natural language priors that capture fine-grained visual characteristics(e.g.,plumage patterns,body proportions,and distinctive markings).These textual descriptions are subsequently used by a conditional diffusion model to synthesize new bird image samples through cross-attention mechanisms that fuse textual embeddings with intermediate visual feature representations during the denoising process,ensuring generated images preserve species-specific morphological details while maintaining photorealistic quality.Additionally,we incorporate a Swin Transformer as the feature extraction backbone whose hierarchical window-based attention mechanism and shifted windowing scheme enable multi-scale local feature extraction that proves particularly effective at capturing finegrained discriminative patterns(such as beak shape and feather texture)while mitigating domain shift between synthetic and original images through consistent feature representations across both data sources.SDNet is validated on both a self-constructed dataset(Bird_BXS)an d a publicly available benchmark(Birds_25),demonstrating substantial improvements over conventional SSL approaches.Our results indicate that the synergistic integration of LLMs,diffusion models,and the Swin Transformer architecture contributes significantly to recognition accuracy,particularly for rare and morphologically similar species.These findings highlight the potential of SDNet for addressing fundamental limitations of existing SSL methods in avian recognition tasks and establishing a new paradigm for efficient self-supervised learning in large-scale ornithological vision applications.
文摘LargeLanguageModels(LLMs)are increasingly appliedinthe fieldof code translation.However,existing evaluation methodologies suffer from two major limitations:(1)the high overlap between test data and pretraining corpora,which introduces significant bias in performance evaluation;and(2)mainstream metrics focus primarily on surface-level accuracy,failing to uncover the underlying factors that constrain model capabilities.To address these issues,this paper presents TCode(Translation-Oriented Code Evaluation benchmark)—a complexity-controllable,contamination-free benchmark dataset for code translation—alongside a dedicated static feature sensitivity evaluation framework.The dataset is carefully designed to control complexity along multiple dimensions—including syntactic nesting and expression intricacy—enabling both broad coverage and fine-grained differentiation of sample difficulty.This design supports precise evaluation of model capabilities across a wide spectrum of translation challenges.The proposed evaluation framework introduces a correlation-driven analysis mechanism based on static program features,enabling predictive modeling of translation success from two perspectives:Code Form Complexity(e.g.,code length and character density)and Semantic Modeling Complexity(e.g.,syntactic depth,control-flow nesting,and type system complexity).Empirical evaluations across representative LLMs—including Qwen2.5-72B and Llama3.3-70B—demonstrate that even state-of-the-art models achieve over 80% compilation success on simple samples,but their accuracy drops sharply below 40% on complex cases.Further correlation analysis indicates that Semantic Modeling Complexity alone is correlated with up to 60% of the variance in translation success,with static program features exhibiting nonlinear threshold effects that highlight clear capability boundaries.This study departs fromthe traditional accuracy-centric evaluation paradigm and,for the first time,systematically characterizes the capabilities of large languagemodels in translation tasks through the lens of programstatic features.The findings provide actionable insights for model refinement and training strategy development.
文摘War rehearsals have become increasingly important in national security due to the growing complexity of international affairs.However,traditional rehearsal methods,such as military chess simulations,are inefficient and inflexible,with particularly pronounced limitations in command and decision-making.The overwhelming volume of information and high decision complexity hinder the realization of autonomous and agile command and control.To address this challenge,an intelligent warfare simulation framework named Command-Agent is proposed,which deeply integrates large language models(LLMs)with digital twin battlefields.By constructing a highly realistic battlefield environment through real-time simulation and multi-source data fusion,the natural language interaction capabilities of LLMs are leveraged to lower the command threshold and to enable autonomous command through the Observe-Orient-Decide-Act(OODA)feedback loop.Within the Command-Agent framework,a multimodel collaborative architecture is further adopted to decouple the decision-generation and command-execution functions of LLMs.By combining specialized models such as Deep Seek-R1 and MCTool,the limitations of single-model capabilities are overcome.MCTool is a lightweight execution model fine-tuned for military Function Calling tasks.The framework also introduces a Vector Knowledge Base to mitigate hallucinations commonly exhibited by LLMs.Experimental results demonstrate that Command-Agent not only enables natural language-driven simulation and control but also deeply understands commander intent.Leveraging the multi-model collaborative architecture,during red-blue UAV confrontations involving 2 to 8 UAVs,the integrated score is improved by an average of 41.8%compared to the single-agent system(MCTool),accompanied by a 161.8%optimization in the battle loss ratio.Furthermore,when compared with multi-agent systems lacking the knowledge base,the inclusion of the Vector Knowledge Base further improves overall performance by 16.8%.In comparison with the general model(Qwen2.5-7B),the fine-tuned MCTool leads by 5%in execution efficiency.Therefore,the proposed Command-Agent introduces a novel perspective to the military command system and offers a feasible solution for intelligent battlefield decision-making.
基金National Key Research and Development Program of China(2024YFC3505400)Capital Clinical Project of Beijing Municipal Science&Technology Commission(Z221100007422092)Capital’s Funds for Health Improvement and Research(2024-1-2231).
文摘Objective To develop a clinical decision and prescription generation system(CDPGS)specifically for diarrhea in traditional Chinese medicine(TCM),utilizing a specialized large language model(LLM),Qwen-TCM-Dia,to standardize diagnostic processes and prescription generation.Methods Two primary datasets were constructed:an evaluation benchmark and a fine-tuning dataset consisting of fundamental diarrhea knowledge,medical records,and chain-ofthought(CoT)reasoning datasets.After an initial evaluation of 16 open-source LLMs across inference time,accuracy,and output quality,Qwen2.5 was selected as the base model due to its superior overall performance.We then employed a two-stage low-rank adaptation(LoRA)fine-tuning strategy,integrating continued pre-training on domain-specific knowledge with instruction fine-tuning using CoT-enriched medical records.This approach was designed to embed the clinical logic(symptoms→pathogenesis→therapeutic principles→prescriptions)into the model’s reasoning capabilities.The resulting fine-tuned model,specialized for TCM diarrhea,was designated as Qwen-TCM-Dia.Model performance was evaluated for disease diagnosis and syndrome type differentiation using accuracy,precision,recall,and F1-score.Furthermore,the quality of the generated prescriptions was compared with that of established open-source TCM LLMs.Results Qwen-TCM-Dia achieved peak performance compared to both the base Qwen2.5 model and five other open-source TCM LLMs.It achieved 97.05%accuracy and 91.48%F1-score in disease diagnosis,and 74.54%accuracy and 74.21%F1-score in syndrome type differentiation.Compared with existing open-source TCM LLMs(BianCang,HuangDi,LingDan,TCMLLM-PR,and ZhongJing),Qwen-TCM-Dia exhibited higher fidelity in reconstructing the“symptoms→pathogenesis→therapeutic principles→prescriptions”logic chain.It provided complete prescriptions,whereas other models often omitted dosages or generated mismatched prescriptions.Conclusion By integrating continued pre-training,CoT reasoning,and a two-stage fine-tuning strategy,this study establishes a CDPGS for diarrhea in TCM.The results demonstrate the synergistic effect of strengthening domain representation through pre-training and activating logical reasoning via CoT.This research not only provides critical technical support for the standardized diagnosis and treatment of diarrhea but also offers a scalable paradigm for the digital inheritance of expert TCM experience and the intelligent transformation of TCM.
基金supported by 2023 Higher Education Scientific Research Planning Project of China Society of Higher Education(No.23PG0408)2023 Philosophy and Social Science Research Programs in Jiangsu Province(No.2023SJSZ0993)+2 种基金Nantong Science and Technology Project(No.JC2023070)Key Project of Jiangsu Province Education Science 14th Five-Year Plan(Grant No.B-b/2024/02/41)the Open Fund of Advanced Cryptography and System Security Key Laboratory of Sichuan Province(Grant No.SKLACSS-202407).
文摘Large language models(LLMs)have revolutionized AI applications across diverse domains.However,their widespread deployment has introduced critical security vulnerabilities,particularly prompt injection attacks that manipulate model behavior through malicious instructions.Following Kitchenham’s guidelines,this systematic review synthesizes 128 peer-reviewed studies from 2022 to 2025 to provide a unified understanding of this rapidly evolving threat landscape.Our findings reveal a swift progression from simple direct injections to sophisticated multimodal attacks,achieving over 90%success rates against unprotected systems.In response,defense mechanisms show varying effectiveness:input preprocessing achieves 60%–80%detection rates and advanced architectural defenses demonstrate up to 95%protection against known patterns,though significant gaps persist against novel attack vectors.We identified 37 distinct defense approaches across three categories,but standardized evaluation frameworks remain limited.Our analysis attributes these vulnerabilities to fundamental LLM architectural limitations,such as the inability to distinguish instructions from data and attention mechanism vulnerabilities.This highlights critical research directions such as formal verification methods,standardized evaluation protocols,and architectural innovations for inherently secure LLM designs.
基金Hebei Province Higher Education Scientific Research Project(QN2025367)Zhangjiakou City 2022 Municipal Science and Technology Plan Self-raised Fund Project(221105D)Hebei Province Education Science“14th Five-Year Plan”Project(2404224).
文摘Objective To develop QingNangTCM,a specialized large language model(LLM)tailored for expert-level traditional Chinese medicine(TCM)question-answering and clinical reasoning,addressing the scarcity of domain-specific corpora and specialized alignment.Methods We constructed QnTCM_Dataset,a corpus of 100000 entries,by integrating data from ShenNong_TCM_Dataset and SymMap v2.0,and synthesizing additional samples via retrieval-augmented generation(RAG)and persona-driven generation.The dataset comprehensively covers diagnostic inquiries,prescriptions,and herbal knowledge.Utilizing P-Tuning v2,we fine-tuned the GLM-4-9B-Chat backbone to develop QingNangTCM.A multidimensional evaluation framework,assessing accuracy,coverage,consistency,safety,professionalism,and fluency,was established using metrics such as bilingual evaluation understudy(BLEU),recall-oriented understudy for gisting evaluation(ROUGE),metric for evaluation of translation with explicit ordering(METEOR),and LLM-as-a-Judge with expert review.Qualitative analysis was conducted across four simulated clinical scenarios:symptom analysis,disease treatment,herb inquiry,and failure cases.Baseline models included GLM-4-9BChat,DeepSeek-V2,HuatuoGPT-II(7B),and GLM-4-9B-Chat(freeze-tuning).Results QingNangTCM achieved the highest scores in BLEU-1/2/3/4(0.425/0.298/0.137/0.064),ROUGE-1/2(0.368/0.157),and METEOR(0.218),demonstrating a balanced and superior normalized performance profile of 0.900 across the dimensions of accuracy,coverage,and consistency.Although its ROUGE-L score(0.299)was lower than that of HuatuoGPT-II(7B)(0.351),it significantly outperformed domain-specific models in expert-validated win rates for professionalism(86%)and safety(73%).Qualitative analysis confirmed that the model strictly adheres to the“symptom-syndrome-pathogenesis-treatment”reasoning chain,though occasional misclassifications and hallucinations persisted when dealing with rare medicinal materials and uncommon syndromes.
基金supported by the Fundamental Research Funds for the Central Universities(No.CUC25SG013)the Foundation of Key Laboratory of Education Informatization for Nationalities(Yunnan Normal University),Ministry of Education(No.EIN2024C006).
文摘Online Public Opinion Reports consolidate news and social media for timely crisis management by governments and enterprises.While large language models(LLMs)enable automated report generation,this specific domain lacks formal task definitions and corresponding benchmarks.To bridge this gap,we define the Automated Online Public Opinion Report Generation(OPOR-Gen)task and construct OPOR-Bench,an event-centric dataset with 463 crisis events across 108 countries(comprising 8.8 K news articles and 185 K tweets).To evaluate report quality,we propose OPOR-Eval,a novel agent-based framework that simulates human expert evaluation.Validation experiments show OPOR-Eval achieves a high Spearman’s correlation(ρ=0.70)with human judgments,though challenges in temporal reasoning persist.This work establishes an initial foundation for advancing automated public opinion reporting research.
基金National Key Research and Development Program of China,Grant/Award Number:2024YFF0507404Special Clinical Business Fund for High-Level Hospitals of China-Japan Friendship Hospital,Grant/Award Number:2024-NHLHCRF-TS-01。
文摘Background:Large language models(LLMs)have shown considerable promise in supporting clinical decision-making.However,their adoption and evaluation in dermatology remains limited.This study aimed to explore the preferences of Chinese dermatologists regarding LLM-generated responses in clinical psoriasis scenarios and to assess how they prioritize key quality dimensions,including accuracy,traceability,and logicality.Methods:A cross-sectional,web-based survey was conducted between December 25,2024,and January 22,2025,following the Checklist for Reporting Results of Internet E-Surveys guidelines.A total of 1247 valid responses were collected from practicing dermatologists across 33 of China's provincial-level administrative divisions.Participants evaluated responses to five categories of clinical questions(etiology,clinical presentation,differential diagnosis,treatment,and case study)generated by five LLMs:ChatGPT-4o,Kimi.ai,Doubao,ZuoYiGPT,and Lingyi-agent.Statistical associations between participant characteristics and model preferences were examined using chi-square tests.Results:ChatGPT-4o(Model 1)emerged as the most preferred model across all clinical tasks,consistently receiving the highest number of votes in case study(n=740),clinical presentation(n=666),differential diagnosis(n=707),etiology(n=602),and treatment(n=656).Significant variation in model preference by professional title was observed only for the differential diagnosis task(χ^(2)=21.13,df=12,p=0.0485),while no significant differences were found across hospital tiers(p>0.05).In terms of evaluation dimensions,accuracy was most frequently rated as“very important”(n=635).A significant association existed between hospital tier and the most valued dimension(χ^(2)=27.667,df=9,p=0.0011),with dermatologists in primary hospitals prioritizing traceability more than their peers in higher-tier hospitals.No significant associations were found across professional titles(p=0.127).Conclusions:Chinese dermatologists suggest a strong preference for ChatGPT-4o over domestic LLMs in psoriasis-related clinical tasks.While accuracy remains the primary criterion,traceability and logicality are also critical,particularly for clinicians in lower-tier hospitals.These findings suggest that future clinical LLMs should prioritize not only content accuracy but also source transparency and structural clarity to meet the diverse needs of different clinical settings.
基金Guangzhou Science and Technology Program,Grant/Award Numbers:2025B03J0110,2024A03J1074,2024A03J0927。
文摘Large language models(LLMs)show considerable potential to revolutionize healthcare through their performance across diverse clinical applications.Given the inherent constraints of LLMs and the critical nature of medical practice,a rigorous and systematic evaluation of their medical competence is imperative.This study presents a comprehensive review of the established methodologies and benchmarks for evaluating the medical competence of LLMs,encompassing a thorough analysis of current assessment practices across medical knowledge,clinical practice competence,and ethical-safety considerations.By integrating clinician competency assessment frameworks into LLMs evaluation,we propose a structured tri-dimensional framework that systematically organizes existing evaluation approaches according to medical theoretical knowledge,clinical practice ability,and ethical-safety considerations.Furthermore,this research provides critical insights into future developmental trajectories while establishing foundational frameworks and standardization protocols for the integration of LLMs into medical practice.
文摘Conversational recommender systems(CRSs)focus on refining preferences and providing personalized recommendations through natural language interactions and dialogue history.Large language models(LLMs)have shown outstanding performance across various domains,thereby prompting researchers to investigate their applicability in recommendation systems.However,due to the lack of task-specific knowledge and an inefficient feature extraction process,LLMs still have suboptimal performance in recommendation tasks.Therefore,external knowledge sources,such as knowledge graphs(KGs)and knowledge bases(KBs),are often introduced to address the issue of data sparsity.Compared to KGs,KBs possess higher retrieval efficiency,making them more suitable for scenarios where LLMs serve as recommenders.To this end,we introduce a novel framework integrating LLMs with KBs for enhanced retrieval generation,namely LLMKB.LLMKB initially leverages structured knowledge to create mapping dictionaries,extracting entity-relation information from heterogeneous knowledge to construct KBs.Then,LLMKB achieves the embedding calibration between user information representations and documents in KBs through retrieval model fine-tuning.Finally,LLMKB employs retrievalaugmented generation to produce recommendations based on fused text inputs,followed by post-processing.Experiment results on two public CRS datasets demonstrate the effectiveness of our framework.Our code is publicly available at the link:https://anonymous.4open.science/r/LLMKB-6FD0.
基金supported by the National Key Research and Development Program of China(Grant No.2024YFA1408604 for K.C.and X.C.)the National Natural Science Foundation of China(Grant Nos.12047503,12447103 for K.C.and X.C.,12325501 for P.Z.,and 12275263 for Y.D.and S.H.)+1 种基金the Innovation Program for Quantum Science and Technology(Grant No.2021ZD0301900 for Y.D.and S.H.)the Natural Science Foundation of Fujian Province of China(Grant No.2023J02032 for Y.D.and S.H.)。
文摘Fundamental physics often confronts complex symbolic problems with few guiding exemplars or established principles.While artificial intelligence(AI)offers promise,its typical need for vast datasets to learn from hinders its use in these information-scarce frontiers.We introduce learning at criticality(LaC),a reinforcement learning scheme that tunes large language models(LLMs)to a sharp learning transition,addressing this information scarcity.At this transition,LLMs achieve peak generalization from minimal data,exemplified by 7-digit base-7 addition-a test of nontrivial arithmetic reasoning.To elucidate this peak,we analyze a minimal concept-network model designed to capture the essence of how LLMs might link tokens.Trained on a single exemplar,this model also undergoes a sharp learning transition.This transition exhibits hallmarks of a second-order phase transition,notably power-law distributed solution path lengths.At this critical point,the system maximizes a“critical thinking pattern”crucial for generalization,enabled by the underlying scale-free exploration.This suggests LLMs reach peak performance by operating at criticality,where such explorative dynamics enable the extraction of underlying operational rules.We demonstrate LaC in quantum field theory:an 8B-parameter LLM,tuned to its critical point by LaC using a few exemplars of symbolic Matsubara sums,solves unseen,higher-order problems,significantly outperforming far larger models.LaC thus leverages critical phenomena,a physical principle,to empower AI for complex,data-sparse challenges in fundamental physics.
文摘We analyze the suitability of existing pre-trained transformer-based language models(PLMs)for abstractive text summarization on German technical healthcare texts.The study focuses on the multilingual capabilities of these models and their ability to perform the task of abstractive text summarization in the healthcare field.The research hypothesis was that large language models could perform high-quality abstractive text summarization on German technical healthcare texts,even if the model is not specifically trained in that language.Through experiments,the research questions explore the performance of transformer language models in dealing with complex syntax constructs,the difference in performance between models trained in English and German,and the impact of translating the source text to English before conducting the summarization.We conducted an evaluation of four PLMs(GPT-3,a translation-based approach also utilizing GPT-3,a German language Model,and a domain-specific bio-medical model approach).The evaluation considered the informativeness using 3 types of metrics based on Recall-Oriented Understudy for Gisting Evaluation(ROUGE)and the quality of results which is manually evaluated considering 5 aspects.The results show that text summarization models could be used in the German healthcare domain and that domain-independent language models achieved the best results.The study proves that text summarization models can simplify the search for pre-existing German knowledge in various domains.
基金Supported by China Health Promotion Foundation Young Doctors’Research Foundation for Inflammatory Bowel DiseaseTaishan Scholars Program of Shandong Province,China,NO.tsqn202306343National Natural Science Foundation of China,No.82270580,No.82070552,No.82270578,and No.82300599.
文摘BACKGROUND Gastrointestinal diseases have complex etiologies and clinical presentations.An accurate diagnosis requires physicians to integrate diverse information,including medical history,laboratory test results,and imaging findings.Existing artificial intelligence-assisted diagnostic tools are limited to single-modality information,resulting in recommendations that are often incomplete and may be associated with clinical or legal risks.AIM To develop and evaluate a collaborative multimodal large language model(LLM)framework for clinical decision-making in digestive diseases.METHODS In this observational study,DeepGut,a multimodal LLM collaborative diagnostic framework,was developed to integrate four distinct large models into a four-tiered structure.The framework sequentially accomplishes multimodal infor-mation extraction,logical“chain”construction,diagnostic and treatment suggestion generation,and risk analysis.The model was evaluated using objective metrics,which assess the reliability and comprehensiveness of model-generated results,and subjective expert opinions,which examine the effectiveness of the framework in assisting physicians.RESULTS The diagnostic and treatment recommendations generated by the DeepGut framework achieved exceptional performance,with a diagnostic accuracy of 97.8%,diagnostic completeness of 93.9%,treatment plan accuracy of 95.2%,and treatment plan completeness of 98.0%,significantly surpassing the capabilities of single-modal LLM-based diagnostic tools.Experts evaluating the framework commended the completeness,relevance,and logical coherence of its outputs.However,the collaborative multimodal LLM approach resulted in increased input and output token counts,leading to higher computational costs and extended diagnostic times.CONCLUSION The framework achieves successful integration of multimodal diagnostic data,demonstrating enhanced performance enabled by multimodal LLM collaboration,which opens new horizons for the clinical application of artificial intelligence-assisted technology.
