Search Results (100)

This study develops a domain-adaptive multimodal RAG (Retrieval-Augmented Generation) system to improve the accuracy and efficiency of technical question answering based on large-scale structured manuals. Using Hyundai Staria maintenance documents as a case study, we extracted text and images from PDF manuals and constructed QA, RAG, and Multi-Turn datasets to reflect realistic troubleshooting scenarios. To overcome limitations of baseline RAG models, we proposed an enhanced architecture that incorporates sentence-level similarity annotations and parameter-efficient fine-tuning via LoRA (Low-Rank Adaptation) using the bLLossom-8B language model and BAAI-bge-m3 embedding model. Experimental results show that the proposed system achieved improvements of 3.0%p in BERTScore, 3.0%p in cosine similarity, and 18.0%p in ROUGE-L compared to existing RAG systems, with notable gains in image-guided response accuracy. A qualitative evaluation by 20 domain experts yielded an average satisfaction score of 4.4 out of 5. This study presents a practical and extensible AI framework for multimodal document understanding, with broad applicability across automotive, industrial, and defense-related technical documentation. Full article

To address issues commonly observed during the inference phase of large language models—such as inconsistent labels, formatting errors, or semantic deviations—a series of targeted strategies has been proposed. First, a relation label refinement strategy based on semantic similarity and syntactic structure has been designed to calibrate the model’s outputs, thereby improving the accuracy and consistency of label prediction. Second, to meet the contextual modeling needs of different types of instance bags, a multi-level contextual augmentation strategy has been constructed. For multi-sentence instance bags, a graph-based retrieval enhancement mechanism is introduced, which integrates intra-bag entity co-occurrence networks with document-level sentence association graphs to strengthen the model’s understanding of cross-sentence semantic relations. For single-sentence instance bags, a semantic expansion strategy based on term frequency-inverse document frequency is employed to retrieve similar sentences. This enriches the training context under the premise of semantic consistency, alleviating the problem of insufficient contextual information. Notably, the proposed multi-granularity framework captures semantic symmetry between entities and relations across different levels of context, which is crucial for accurate and balanced relation understanding. The proposed methodology offers practical advancements for semantic analysis applications, particularly in knowledge graph development. Full article

Customer churn prediction has become crucial for businesses, yet it poses significant challenges regarding privacy preservation and prediction accuracy. In this paper, we address two fundamental questions: (1) How can customer churn be effectively predicted while ensuring robust privacy protection of sensitive data? (2) How can large language models enhance churn prediction accuracy while maintaining data privacy? To address these questions, we propose FairRAG, a robust architecture that combines differential privacy, retrieval-augmented generation, and LLMs. Our approach leverages OPT-125M as the core language model along with a sentence transformer for semantic similarity matching while incorporating differential privacy mechanisms to generate synthetic training data. We evaluate FairRAG on two diverse datasets: Bank Churn and Telco Churn. The results demonstrate significant improvements over both traditional machine learning approaches and standalone LLMs, achieving accuracy improvements of up to 11% on the Bank Churn dataset and 12% on the Telco Churn dataset. These improvements were maintained when using differentially private synthetic data, thus indicating robust privacy and accuracy trade-offs. Full article

Career planning agencies and other organizations can help workers if they are able to effectively identify related occupations that are relevant to the task at hand. Occupational knowledge bases such as O*NET and ESCO represent mature attempts to categorize occupations and describe them in detail so that they can be used to search for related occupations. Vector databases offer an opportunity to find related occupations based on large pre-trained word and sentence embeddings and their associated retrieval algorithms for similarity search. This paper reports a systematic empirical evaluation of the possibilities of using vector databases for related occupation retrieval using different document structures, embeddings, and retrieval configurations for two popular open source vector databases, and using the O*NET curated database. The objective was to understand the extent to which curated relations capture all the meaningful relations in a context of retrieval. The results show that, independent of the database used, distance metrics, sentence embeddings, and the selection of text fragments are all significant in the overall retrieval performance when comparing with curated relations, but they also retrieve other relevant occupations based on text similarity. Further, the precision is high for smaller cutoffs in the results list, which is especially important for settings in which vector database retrieval is set up as part of a Retrieval Augmented Generation (RAG) pattern. The inspection of highly ranked retrieved related occupations not explicit in the curated database reveals that text similarity captures the taxonomical grouping of some occupations in some cases, but also other cross-cuts different aspects that are distinct from the hierarchical organization of the database in most of the cases. This suggests that text retrieval should be combined with querying explicit relations in practical applications. Full article

This article introduces a freely available Spanish–Mixtec parallel corpus designed to foster natural language processing (NLP) development for an indigenous language that remains digitally low-resourced. The dataset, comprising 14,587 sentence pairs, covers Mixtec variants from Guerrero (Tlacoachistlahuaca, Northern Guerrero, and Xochapa) and Oaxaca (Western Coast, Southern Lowland, Santa María Yosoyúa, Central, Lower Cañada, Western Central, San Antonio Huitepec, Upper Western, and Southwestern Central). Texts are classified into four main domains as follows: education, law, health, and religion. To compile these data, we conducted a two-phase collection process as follows: first, an online search of government portals, religious organizations, and Mixtec language blogs; and second, an on-site retrieval of physical texts from the library of the Autonomous University of Querétaro. Scanning and optical character recognition were then performed to digitize physical materials, followed by manual correction to fix character misreadings and remove duplicates or irrelevant segments. We conducted a preliminary evaluation of the collected data to validate its usability in automatic translation systems. From Spanish to Mixtec, a fine-tuned GPT-4o-mini model yielded a BLEU score of 0.22 and a TER score of 122.86, while two fine-tuned open source models mBART-50 and M2M-100 yielded BLEU scores of 4.2 and 2.63 and TER scores of 98.99 and 104.87, respectively. All code demonstrating data usage, along with the final corpus itself, is publicly accessible via GitHub and Figshare. We anticipate that this resource will enable further research into machine translation, speech recognition, and other NLP applications while contributing to the broader goal of preserving and revitalizing the Mixtec language. Full article

At present, researchers are showing a marked interest in the topic of few-shot named entity recognition (NER). Previous studies have demonstrated that prompt-based learning methods can effectively improve the performance of few-shot NER models and can reduce the need for annotated data. However, the contextual information of the relationship between core entities and a given prompt may not have been considered in these studies; moreover, research in this field continues to suffer from the negative impact of a limited amount of annotated data. A multi-class label prompt selection and core entity replacement-based named entity recognition (MPSCER-NER) model is proposed in this study. A multi-class label prompt selection strategy is presented, which can assist in the task of sentence–word representation. A long-distance dependency is formed between the sentence and the multi-class label prompt. A core entity replacement strategy is presented, which can enrich the word vectors of training data. In addition, a weighted random algorithm is used to retrieve the core entities that are to be replaced from the multi-class label prompt. The experimental results show that, when implemented on the CoNLL-2003, Ontonotes 5.0, Ontonotes 4.0, and BC5CDR datasets under 5-Way k-Shot (k = 5, 10), the MPSCER-NER model achieves minimum F1-score improvements of 1.32%, 2.14%, 1.05%, 1.32%, 0.84%, 1.46%, 1.43%, and 1.11% in comparison with NNshot, StructShot, MatchingCNN, ProtoBERT, DNER, and SRNER, respectively. Full article

Text-based image retrieval is one of the most common approaches for searching images acquired from vision sensors such as cameras. However, this method suffers from limitations in retrieval accuracy, particularly when the query contains limited information or involves previously unseen sentences. These challenges arise because keyword-based matching fails to adequately capture contextual and semantic meanings. To address these limitations, we propose a novel approach that transforms sentences and images into semantic graphs and scene graphs, enabling a quantitative comparison between them. Specifically, we utilize a graph neural network (GNN) to learn features of nodes and edges and generate graph embeddings, enabling image retrieval through natural language queries without relying on additional image metadata. We introduce a contrastive GNN-based framework that matches semantic graphs with scene graphs to retrieve semantically similar images. In addition, we incorporate a hard negative mining strategy, allowing the model to effectively learn from more challenging negative samples. The experimental results on the Visual Genome dataset show that the proposed method achieves a top nDCG@50 score of 0.745, improving retrieval performance by approximately 7.7 percentage points compared to random sampling with full graphs. This confirms that the model effectively retrieves semantically relevant images by structurally interpreting complex scenes. Full article

Nowadays, accessibility to academic papers has been significantly improved with electric publications on the internet, where open access has become common. At the same time, it has increased workloads in literature surveys for researchers who usually manually download PDF files and check their contents. To solve this drawback, we have proposed a reference paper collection system using a web scraping technology and natural language models. However, our previous system often finds a limited number of relevant reference papers after taking long time, since it relies on one paper search website and runs on a single thread at a multi-core CPU. In this paper, we present an improved reference paper collection system with three enhancements to solve them: (1) integrating the APIs from multiple paper search web sites, namely, the bulk search endpoint in the Semantic Scholar API, the article search endpoint in the DOAJ API, and the search and fetch endpoint in the PubMed API to retrieve article metadata, (2) running the program on multiple threads for multi-core CPU, and (3) implementing Dynamic URL Redirection, Regex-based URL Parsing, and HTML Scraping with URL Extraction for fast checking of PDF file accessibility, along with sentence embedding to assess relevance based on semantic similarity. For evaluations, we compare the number of obtained reference papers and the response time between the proposal, our previous work, and common literature search tools in five reference paper queries. The results show that the proposal increases the number of relevant reference papers by 64.38% and reduces the time by 59.78% on average compared to our previous work, while outperforming common literature search tools in reference papers. Thus, the effectiveness of the proposed system has been demonstrated in our experiments. Full article

With the rapid advancement of human–machine dialogue technology, sentiment analysis has become increasingly crucial. However, deep learning-based methods struggle with interpretability and reliability due to the subjectivity of emotions and the challenge of capturing emotion–cause relationships. To address these issues, we propose a novel sentiment analysis framework that integrates structured commonsense knowledge to explicitly infer emotional causes, enabling causal reasoning between historical and target sentences. Additionally, we enhance sentiment classification by leveraging large language models (LLMs) with dynamic example retrieval, constructing an experience database to guide the model using contextually relevant instances. To further improve adaptability, we design a semantic interpretation task for refining emotion category representations and fine-tune the LLM accordingly. Experiments on three benchmark datasets show that our approach significantly improves accuracy and reliability, surpassing traditional deep-learning methods. These findings underscore the effectiveness of structured reasoning, knowledge retrieval, and LLM-driven sentiment adaptation in advancing emotion–cause-based sentiment analysis. Full article

We present an automated journal recommendation pipeline designed to evaluate the performance of five Sentence Transformer models—all-mpnet-base-v2 (Mpnet), all-MiniLM-L6-v2 (Minilm-l6), all-MiniLM-L12-v2 (Minilm-l12), multi-qa-distilbert-cos-v1 (Multi-qa-distilbert), and all-distilroberta-v1 (roberta)—for recommending journals aligned with a manuscript’s thematic scope. The pipeline extracts domain-relevant keywords from a manuscript via KeyBERT, retrieves potentially related articles from PubMed, and encodes both the test manuscript and retrieved articles into high-dimensional embeddings. By computing cosine similarity, it ranks relevant journals based on thematic overlap. Evaluations on 50 test articles highlight mpnet’s strong performance (mean similarity score 0.71 ± 0.04), albeit with higher computational demands. Minilm-l12 and minilm-l6 offer comparable precision at lower cost, while multi-qa-distilbert and roberta yield broader recommendations better suited to interdisciplinary research. These findings underscore key trade-offs among embedding models and demonstrate how they can provide interpretable, data-driven insights to guide journal selection across varied research contexts. Full article

Empathetic and coherent responses are critical in automated chatbot-facilitated psychotherapy. This study addresses the challenge of enhancing the emotional and contextual understanding of large language models (LLMs) in psychiatric applications. We introduce Emotion-Aware Embedding Fusion, a novel framework integrating hierarchical fusion and attention mechanisms to prioritize semantic and emotional features in therapy transcripts. Our approach combines multiple emotion lexicons, including NRC Emotion Lexicon, VADER, WordNet, and SentiWordNet, with state-of-the-art LLMs such as Flan-T5, Llama 2, DeepSeek-R1, and ChatGPT 4. Therapy session transcripts, comprising over 2000 samples, are segmented into hierarchical levels (word, sentence, and session) using neural networks, while hierarchical fusion combines these features with pooling techniques to refine emotional representations. Attention mechanisms, including multi-head self-attention and cross-attention, further prioritize emotional and contextual features, enabling the temporal modeling of emotional shifts across sessions. The processed embeddings, computed using BERT, GPT-3, and RoBERTa, are stored in the Facebook AI similarity search vector database, which enables efficient similarity search and clustering across dense vector spaces. Upon user queries, relevant segments are retrieved and provided as context to LLMs, enhancing their ability to generate empathetic and contextually relevant responses. The proposed framework is evaluated across multiple practical use cases to demonstrate real-world applicability, including AI-driven therapy chatbots. The system can be integrated into existing mental health platforms to generate personalized responses based on retrieved therapy session data. The experimental results show that our framework enhances empathy, coherence, informativeness, and fluency, surpassing baseline models while improving LLMs’ emotional intelligence and contextual adaptability for psychotherapy. Full article

Vision–language pre-training (VLP) faces challenges in aligning hierarchical textual semantics (words/phrases/sentences) with multi-scale visual features (objects/relations/global context). We propose a hierarchical VLP model (HieVLP) that addresses such challenges through semantic decomposition and progressive alignment. Textually, a semantic parser deconstructs captions into word-, phrase-, and sentence-level components, which are encoded via hierarchical BERT layers. Visually, a Swin Transformer extracts object- (local), relation- (mid-scale), and global-level features through shifted window hierarchies. During pre-training, a freezing strategy sequentially activates text layers (sentence→phrase→word), aligning each with the corresponding visual scales via contrastive and language modeling losses. The experimental evaluations demonstrate that HieVLP outperforms hierarchical baselines across various tasks, with the performance improvements ranging from approximately 3.2% to 11.2%. In the image captioning task, HieVLP exhibits an average CIDEr improvement of around 7.2% and a 2.1% improvement in the SPICE metric. For image–text retrieval, it achieves recall increases of 4.7–6.8%. In reasoning tasks, HieVLP boosts accuracy by 2.96–5.8%. These results validate that explicit multi-level alignment enables contextually coherent caption generation and precise cross-modal reasoning. Full article

Identifying key concepts in automobile fault texts is crucial for understanding fault causes and enabling diagnosis. However, effective mining tools are lacking, leaving much latent information unexplored. To solve the problem, this paper proposes Chinese named entity recognition for automobile fault texts based on external context retrieval and adversarial training. First, we retrieve external contexts by using a search engine. Then, the input sentence and its external contexts are respectively fed into Lexicon Enhanced BERT to improve the text embedding representation. Furthermore, the input sentence and its external contexts embedding representation are fused through the attention mechanism. Then, adversarial samples are generated by adding perturbations to the fusion vector representation. Finally, the fusion vector representation and adversarial samples are input into the BiLSTM-CRF layer as training data for entity labeling. Our model is evaluated on the automotive fault datasets, Weibo and Resume datasets, and achieves state-of-the-art results. Full article

Current clinical care relies heavily on complex, rule-based systems for tasks like diagnosis and treatment. However, these systems can be cumbersome and require constant updates. This study explores the potential of the large language model (LLM), LLaMA 2, to address these limitations. We tested LLaMA 2′s performance in interpreting complex clinical process models, such as Mayo Clinic Care Pathway Models (CPMs), and providing accurate clinical recommendations. LLM was trained on encoded pathways versions using DOT language, embedding them with SentenceTransformer, and then presented with hypothetical patient cases. We compared the token-level accuracy between LLM output and the ground truth by measuring both node and edge accuracy. LLaMA 2 accurately retrieved the diagnosis, suggested further evaluation, and delivered appropriate management steps, all based on the pathways. The average node accuracy across the different pathways was 0.91 (SD ± 0.045), while the average edge accuracy was 0.92 (SD ± 0.122). This study highlights the potential of LLMs for healthcare information retrieval, especially when relevant data are provided. Future research should focus on improving these models’ interpretability and their integration into existing clinical workflows. Full article

Aspect category sentiment analysis aims to automatically identify the sentiment polarities of aspect categories mentioned in text, and is widely used in the data analysis of product reviews and social media. Most existing studies typically limit themselves to utilizing sentence-level local information, thereby failing to fully exploit the potential of document-level and corpus-level global information. To address these limitations, we propose a model that integrates global information for aspect category sentiment analysis, aiming to leverage sentence-level, document-level, and corpus-level information simultaneously. Specifically, based on sentences and their corresponding aspect categories, a graph neural network is initially built to capture document-level information, including sentiment consistency within the same category and sentiment similarity between different categories in a review. We subsequently employ a memory network to retain corpus-level information, where the representations of training instances serve as keys and their associated labels as values. Additionally, a k-nearest neighbor retrieval mechanism is used to find training instances relevant to a given input. Experimental results on four commonly used datasets from SemEval 2015 and 2016 demonstrate the effectiveness of our model. The in-depth experimental analysis reveals that incorporating document-level information substantially improves the accuracies of the two primary ‘positive’ and ‘negative’ categories, while the inclusion of corpus-level information is especially advantageous for identifying the less frequently occurring ‘neutral’ category. Full article