Search Results (78)

The rapid growth of smart home technologies, driven by the expansion of the Internet of Things (IoT), has introduced both opportunities and challenges in automating daily routines and orchestrating device interactions. Traditional rule-based automation systems often fall short in adapting to dynamic conditions, integrating heterogeneous devices, and responding to evolving user needs. To address these limitations, this study introduces a novel smart home orchestration framework that combines generative Artificial Intelligence (AI), Retrieval-Augmented Generation (RAG), and the modular OSGi framework. The proposed system allows users to express requirements in natural language, which are then interpreted and transformed into executable service bundles by large language models (LLMs) enhanced with contextual knowledge retrieved from vector databases. These AI-generated service bundles are dynamically deployed via OSGi, enabling real-time service adaptation without system downtime. Manufacturer-provided device capabilities are seamlessly integrated into the orchestration pipeline, ensuring compatibility and extensibility. The framework was validated through multiple use-case scenarios involving dynamic device discovery, on-demand code generation, and adaptive orchestration based on user preferences. Results highlight the system’s ability to enhance automation efficiency, personalization, and resilience. This work demonstrates the feasibility and advantages of AI-driven orchestration in realising intelligent, flexible, and scalable smart home environments. Full article

Legal text summarization presents distinct challenges due to the intricate and domain-specific nature of legal language. This paper introduces a novel framework integrating dynamic Retrieval-Augmented Generation (RAG) with domain-specific adaptation to enhance the accuracy and contextual relevance of legal document summaries. The proposed Dynamic Legal RAG system achieves a vital form of symmetry between information retrieval and content generation, ensuring that retrieved legal knowledge is both comprehensive and precise. Using the BM25 retriever with top-3 chunk selection, the system optimizes relevance and efficiency, minimizing redundancy while maximizing legally pertinent content. with top-3 chunk selection, the system optimizes relevance and efficiency, minimizing redundancy while maximizing legally pertinent content. A key design feature is the compression ratio constraint (0.05 to 0.5), maintaining structural symmetry between the original judgment and its summary by balancing representation and information density. Extensive evaluations establish BM25 as the most effective retriever, striking an optimal balance between precision and recall. A comparative analysis of transformer-based (Decoder-only) models—DeepSeek-7B, LLaMA 2-7B, and LLaMA 3.1-8B—demonstrates that LLaMA 3.1-8B, enriched with Legal Named Entity Recognition (NER) and the Dynamic RAG system, achieves superior performance with a BERTScore of 0.89. This study lays a strong foundation for future research in hybrid retrieval models, adaptive chunking strategies, and legal-specific evaluation metrics, with practical implications for case law analysis and automated legal drafting. Full article

Effective and sustainable water reserve management faces increasing challenges due to climate-induced variability, data fragmentation, and the limitations of traditional, static modeling systems. This study introduces a conceptual framework designed to address these challenges by integrating digital twins, IoT-driven real-time monitoring, game engine simulations, and AI-driven decision support systems (AI-DSS). The methodology involves constructing a digital twin ecosystem using IoT sensors, GIS layers, remote-sensing imagery, and game engines. This ecosystem simulates water dynamics and assesses policy interventions in real time. AI components, including machine-learning models and retrieval-augmented generation (RAG) chatbots, are embedded to synthesize real-time data into actionable insights. The framework enables the continuous assessment of hydrological dynamics, predictive risk analysis, and immersive, scenario-based decision-making to support long-term water sustainability. Simulated scenarios demonstrate accurate flood forecasting under variable rainfall intensities, early drought detection based on soil moisture and flow data, and real-time water-quality alerts. Digital elevation models from UAV photogrammetry enhance terrain realism, and AI models support dynamic predictions. Results show how the framework supports proactive mitigation planning, climate adaptation, and stakeholder communication in pursuit of resilient and sustainable water governance. By enabling early intervention, efficient resource allocation, and participatory decision-making, the proposed system fosters long-term, sustainable water security and environmental resilience. This conceptual framework suggests a pathway toward more transparent, data-informed, and resilient decision-making processes in water reserves management, particularly in regions facing climatic uncertainty and infrastructure limitations, aligning with global sustainability goals and adaptive water governance strategies. Full article

Integrating causal inference and estimation methods, especially in Natural Language Processsing (NLP), is essential to improve interpretability and robustness in deep learning (DL) models. The objectives are to present the IV-NLP methodology and its application. IV-NLP integrates two approaches. The first defines the process of the inference and estimation of the causal effect in original, predicted, and synthetic data. The second one includes a validation method of the results obtained by the selected Large-Language Model (LLM). IV-NLP proposes to use synthetic data in predictive tasks only if the causal effect pattern of the synthetic data is aligned with the causal effect pattern of the original data. DL models, the Instrumental Variable (IV) method, statistical methods, and GPT-3.5-turbo-0125 were used for its application, including an intervention method using a variation of the Retrieval-Augmented Generation (RAG) technique. Our findings reveal notable discrepancies between the original and synthetic data, highlighting that the synthetic data do not fully capture the underlying causal effect patterns of the original data, evidencing homogeneity and low diversity in the synthetic data. Interestingly, when evaluating the causal effect in the predictions made by our three best DL models, it was verified that the model with the lowest accuracy (84.50%) was fully aligned with the overall causal effect pattern. These results demonstrate the potential of integrating DL and LLM models with causal inference methods. Full article

Recent advancements in retrieval-augmented generation (RAG) have substantially enhanced the efficiency of information retrieval. However, traditional RAG-based systems still encounter challenges, such as high latency in output decision making, the inaccurate retrieval of road traffic-related laws and regulations, and considerable processing overhead in large-scale searches. This study presents an innovative application of RAG technology for processing road traffic-related laws and regulations, particularly in the context of unmanned systems like autonomous driving. Our approach integrates embedding generation using a LoRA-enhanced BERT-based uncased model and an optimized retrieval strategy that combines maximal marginal similarity score thresholding with contextual compression retrieval. The proposed system enhances and achieves improved retrieval accuracy while reducing processing overhead. Leveraging road traffic-related regulatory datasets, the LoRA-enhanced model demonstrated remarkable performance gains over traditional RAG methods. Specifically, our model reduced the number of trainable parameters by 13.6% and lowered computational costs by 18.7%. Performance evaluations using BLEU, CIDEr, and SPICE scores revealed a 4.36% increase in BLEU-4, a 6.83% improvement in CIDEr, and a 5.46% improved in SPICE, confirming greater structural accuracy in regulatory text generation. Additionally, our method achieved an 8.5% improvement in retrieval accuracy across key metrics, outperforming baseline RAG systems. These contributions pave the way for more efficient and reliable traffic regulation processing, enabling better decision making in autonomous systems. Full article

Retrieval-Augmented Generation (RAG) overcomes the main barrier for the adoption of LLM-based chatbots in education: hallucinations. The uncomplicated architecture of RAG chatbots makes it relatively easy to implement chatbots that serve specific purposes and thus are capable of addressing various needs in the educational domain. With five years having passed since the introduction of RAG, the time has come to check the progress attained in its adoption in education. This paper identifies 47 papers dedicated to RAG chatbots’ uses for various kinds of educational purposes, which are analyzed in terms of their character, the target of the support provided by the chatbots, the thematic scope of the knowledge accessible via the chatbots, the underlying large language model, and the character of their evaluation. Full article

Similar patent document retrieval is an essential task that reduces the scope of patent claimants’ searches, and numerous studies have attempted to provide automated patent search services. Recently, Retrieval-Augmented Generation (RAG) based on generative language models has emerged as an excellent method for accessing and utilizing patent knowledge environments. RAG-based patent search services offer enhanced retrieval ranking performance as AI search services by providing document knowledge similar to queries. However, achieving optimal similarity-based document ranking in search services remains a challenging task, as search methods based on document similarity do not adequately address the characteristics of patent documents. Unlike general document retrieval, the similarity of patent documents must take into account prior art relationships. To address this issue, we propose PAI-NET, a deep neural network for computing patent document similarities by incorporating expert knowledge of prior art relationships. We demonstrate that our proposed method outperforms current state-of-the-art models in patent document classification tasks through semantic distance evaluation on the USPD and KPRIS datasets. PAI-NET presents similar document candidates, demonstrating a superior patent search performance improvement of 15% over state-of-the-art methods. Full article

In recent years, the rapid development of the new energy vehicle (NEV) industry has exposed significant deficiencies in intelligent fault diagnosis and information retrieval technologies, especially in intelligent fault information retrieval, which faces persistent challenges including inadequate system adaptability and reasoning bottlenecks. To address these challenges, this study proposes a Retrieval-Augmented Generation (RAG) framework that integrates large language models (LLMs) with knowledge graphs (KGs). The framework consists of three key components: fault data collection, knowledge graph construction, and fault knowledge model training. The primary research contributions are threefold: (1) A domain-optimized fine-tuning strategy for LLMs based on NEV fault characteristics, verifying the superior accuracy of the Bidirectional Encoder Representations from Transformers (BERT) model in fault classification tasks. (2) A structured knowledge graph encompassing 122 fault categories, developed through the ChatGLM3-6B model completing named entity and knowledge relation extraction to generate fault knowledge and build a paraphrased vocabulary. (3) An intelligent fault information retrieval system that significantly outperforms traditional models in NEV-specific Q&A scenarios, providing multi-level fault cause analysis and actionable solution recommendations. Full article

This study investigates prompt engineering (PE) strategies to mitigate hallucination, a key limitation of multimodal large language models (MLLMs). To address this issue, we explore five prominent multimodal PE techniques: in-context learning (ICL), chain of thought (CoT), step-by-step reasoning (SSR), tree of thought (ToT), and retrieval-augmented generation (RAG). These techniques are systematically applied across multiple datasets with distinct domains and characteristics. Based on the empirical findings, we propose the greedy prompt engineering strategy (Greedy PES), a methodology for optimizing PE application across different datasets and MLLM models. To evaluate user satisfaction with MLLM-generated responses, we adopt a comprehensive set of evaluation metrics, including BLEU, ROUGE, METEOR, S-BERT, MoverScore, and CIDEr. A weighted aggregate evaluation score is introduced to provide a holistic assessment of model performance under varying conditions. Experimental results demonstrate that the optimal prompt engineering strategy varies significantly depending on both dataset properties and the MLLM model used. Specifically, datasets categorized as general benefit the most from ICL, ToT, and RAG, whereas mathematical datasets perform optimally with ICL, SSR, and ToT. In scientific reasoning tasks, RAG and SSR emerge as the most effective strategies. Applying Greedy PES leads to a substantial improvement in performance across different multimodal tasks, achieving an average evaluation score enhancement of 184.3% for general image captioning, 90.3% for mathematical visual question answering (VQA), and 49.1% for science visual question answering (VQA) compared to conventional approaches. These findings highlight the effectiveness of structured PE strategies in optimizing MLLM performance and provide a robust framework for PE-driven model enhancement across diverse multimodal applications. Full article

Knee osteoarthritis (KOA) is a prevalent condition globally, leading to significant pain and disability, particularly in individuals over the age of 40. While exercise has been shown to reduce symptoms and improve physical function and quality of life in patients with KOA, long-term adherence to exercise programs remains a challenge due to the lack of ongoing support. To address this, a chatbot was developed using large language models (LLMs) to provide evidence-based guidance and promote adherence to treatment. A systematic review conducted under the PRISMA framework identified relevant clinical guidelines that served as the foundational knowledge base for the chatbot. The Mistral 7B model, optimized with Parameter-Efficient Fine-Tuning (PEFT) and Mixture-of-Experts (MoE) techniques, was integrated to ensure computational efficiency and mitigate hallucinations, a critical concern in medical applications. Additionally, the chatbot employs Self-Reflective Retrieval-Augmented Generation (SELF-RAG) combined with Chain of Thought (CoT) reasoning, enabling dynamic query reformulation and the generation of accurate, evidence-based responses tailored to patient needs. The chatbot was evaluated by comparing pre- and post-improvement versions and against a reference model (ChatGPT), using metrics of accuracy, relevance, and consistency. The results demonstrated significant improvements in response quality and conversational coherence, emphasizing the potential of integrating advanced LLMs with retrieval and reasoning methods to address critical challenges in healthcare. This approach not only enhances treatment adherence but also strengthens patient–provider interactions in managing chronic conditions like KOA. Full article

Background: Large language models (LLMs) have the potential to enhance information processing and clinical reasoning in the healthcare industry but are hindered by inaccuracies and hallucinations. The retrieval-augmented generation (RAG) technique may address these problems by integrating external knowledge sources. Methods: We developed a RAG-based chatbot called Thyro-GenAI by integrating a database of textbooks and guidelines with LLM. Thyro-GenAI and three service LLMs: OpenAI’s ChatGPT-4o, Perplexity AI’s ChatGPT-4o, and Anthropic’s Claude 3.5 Sonnet, were asked personalized clinical questions about thyroid disease. Three thyroid specialists assessed the quality of the generated responses and references without being blinded, which allowed them to interact with different chatbot interfaces. Results: Thyro-GenAI achieved the highest inverse-weighted mean rank for overall response quality. The overall inverse-weighted mean rankings for Thyro-GenAI, ChatGPT, Perplexity, and Claude were 3.0, 2.3, 2.8, and 1.9, respectively. Thyro-GenAI also achieved the second-highest inverse-weighted mean rank for overall reference quality. The overall inverse-weighted mean rankings for Thyro-GenAI, ChatGPT, Perplexity, and Claude were 3.1, 2.3, 3.2, and 1.8, respectively. Conclusions: Thyro-GenAI produced patient-specific clinical reasoning output based on a vector database, with fewer hallucinations and more reliability, compared to service LLMs. This emphasis on evidence-based responses ensures its safety and validity, addressing a critical limitation of existing LLMs. By integrating RAG with LLMs, it has the potential to support frontline clinical decision-making, especially helping first-line physicians by offering reliable decision support while managing thyroid disease patients. Full article

With the increasing diversification of ore types and the complexity of processing techniques in the mining industry, traditional decision-making methods for mineral processing flowsheets can no longer meet the high efficiency and intelligence requirements. This paper proposes a knowledge graph-based framework for constructing a mineral-processing design knowledge base and knowledge reasoning, aiming at providing intelligent and efficient decision support for mining engineers. This framework integrates Chinese NLP models for text vectorization, optimizes prompt generation through Retrieval Augmented Generation (RAG) technology, realizes knowledge graph construction, and implements knowledge reasoning for nonferrous metal mineral-processing design using large reasoning models. By analyzing the genetic characteristics of ores and the requirements of processing techniques, the framework outputs reasonable flowsheet designs, which could help engineers save research time and labor in optimizing processes, selecting suitable reagents, and adjusting process parameters. Through decision analysis of the mineral-processing flowsheets for three typical copper mines, the framework demonstrates its advantages in improving process flowsheet design, and shows good potential for further application in complex mineral-processing environments. Full article

To enhance the accessibility and accuracy of plant protection knowledge for agricultural practitioners, this study develops an intelligent question-answering (QA) system based on a large language model (LLM). A local knowledge base was constructed by vectorizing 7000 research papers and books in the field of plant protection, from which 568 representative papers were selected to generate QA data using an LLM. After data cleaning and filtering, a fine-tuning dataset comprising 9000 question–answer pairs was curated. To optimize the model’s performance, low-rank adaptation (LoRA) was applied to the InterLM-20B model, resulting in the InterLM-20B-LoRA, which was integrated with Langchain-ChatChat and the local knowledge base to develop the QA system. Additionally, retrieval-augmented generation (RAG) technology was implemented to enhance response accuracy by enabling the model to retrieve relevant field-specific knowledge before generating answers, effectively mitigating the risk of hallucinated information. The experimental results demonstrate that the proposed system achieves an answer accuracy of 97%, highlighting its potential as an advanced solution for intelligent agricultural QA services. Full article

Graph-based retrieval-augmented generation (GraphRAG) represents an innovative advancement in natural language processing, leveraging the power of large language models (LLMs) for complex tasks such as policy generation. This research presents a GraphRAG model trained on YouTube data containing keywords related to logistics issues to generate policy proposals addressing these challenges. The collected data include both video subtitles and user comments, which are used to fine-tune the GraphRAG model. To evaluate the effectiveness of this approach, the performance of the proposed model is compared to a standard generative pre-trained transformer (GPT) model. The results show that the GraphRAG model outperforms the GPT model in most prompts, highlighting its potential to generate more accurate and contextually relevant policy recommendations. This study not only contributes to the evolving field of LLM-based natural language processing (NLP) applications but also explores new methods for improving model efficiency and scalability in real-world domains like logistics policy making. Full article

The Internet of Drones (IoD) integrates autonomous aerial platforms with security, logistics, agriculture, and disaster relief. Decision-making in IoD suffers in real-time adaptability, platform interoperability, and scalability. Conventional decision frameworks with heuristic algorithms and narrow Artificial Intelligence (AI) falter in complex environments. To mitigate these, in this study, an augmented decision model is proposed, combining large language models (LLMs) and retrieval-augmented generation (RAG) for enhancing IoD intelligence. Centralized intelligence is achieved by processing environment factors, mission logs, and telemetry, with real-time adaptability. Efficient retrieval of contextual information through RAG is merged with LLMs for timely, correct decision-making. Contextualized decision-making vastly improves adaptability in uncertain environments for a drone network. With LLMs and RAG, the model introduces a scalable, adaptable IoD operations solution. It enables the development of autonomous aerial platforms in industries, with future work in computational efficiency, ethics, and extending operational environments. In-depth analysis with the collection of drone telemetry logs and operational factors was conducted. Decision accuracy, response time, and contextual relevance were measured to gauge system effectiveness. The model’s performance increased remarkably, with a BLEU of 0.82 and a cosine similarity of 0.87, proving its effectiveness for operational commands. Decision latency averaged 120 milliseconds, proving its suitability for real-time IoD use cases. Full article