Named Entity Recognition Research Articles

COMCARE: A Collaborative Ensemble Framework for Context-Aware Medical Named Entity Recognition and Relation Extraction

The rapid expansion of medical information has resulted in named entity recognition (NER) and relation extraction (RE) essential for clinical decision support systems. Medical texts often contain specialized vocabulary, ambiguous abbreviations, synonyms, polysemous terms, and overlapping entities, which introduce significant challenges to the extraction process. Existing approaches, which typically rely on single models such as BiLSTM or BERT, often struggle with these complexities. Although large language models (LLMs) have shown promise in various NLP tasks, they still face limitations in handling token-level tasks critical for medical NER and RE. To address these challenges, we propose COMCARE, a collaborative ensemble framework for context-aware medical NER and RE that integrates multiple pre-trained language models through a collaborative decision strategy. For NER, we combined PubMedBERT and PubMed-T5, leveraging PubMedBERT’s contextual understanding and PubMed-T5’s generative capabilities to handle diverse forms of medical terminology, from standard domain-specific jargon to nonstandard representations, such as uncommon abbreviations and out-of-vocabulary (OOV) terms. For RE, we integrated general-domain BERT with biomedical-specific BERT and PubMed-T5, utilizing token-level information from the NER module to enhance the context-aware entity-based relation extraction. To effectively handle long-range dependencies and maintain consistent performance across diverse texts, we implemented a semantic chunking approach and combined the model outputs through a majority voting mechanism. We evaluated COMCARE on several biomedical datasets, including BioRED, ADE, RDD, and DIANN Corpus. For BioRED, COMCARE achieved F1 scores of 93.76% for NER and 68.73% for RE, outperforming BioBERT by 1.25% and 1.74%, respectively. On the RDD Corpus, COMCARE showed F1 scores of 77.86% for NER and 86.79% for RE while achieving 82.48% for NER on ADE and 99.36% for NER on DIANN. These results demonstrate the effectiveness of our approach in handling complex medical terminology and overlapping entities, highlighting its potential to improve clinical decision support systems.

Clinical entity-aware domain adaptation in low resource setting for inflammatory bowel disease

The digitization of healthcare records has revolutionized medical research and patient care, with electronic health records (EHRs) containing a wealth of structured and unstructured data. Extracting valuable information from unstructured clinical text presents a significant challenge, necessitating automated tools for efficient data mining. Natural language processing (NLP) methods have been pivotal in this endeavor, aiming to extract crucial clinical concepts embedded within free-form text. Our research addresses the imperative for robust biomedical entity extraction, focusing specifically on inflammatory bowel disease (IBD). Leveraging novel domain-specific pre-training and entity-aware masking strategies with contrastive learning, we fine-tune and adapt a general language model to be better adapted to IBD-related information extraction scenarios. Our named entity recognition (NER) tool streamlines the retrieval process, supporting annotation, correction, and visualization functionalities. In summary, we developed a comprehensive pipeline for clinical Dutch NER encompassing an efficient domain adaptation strategy with domain-aware masking and model fine-tuning enhancements, and an end-to-end entity extraction tool, significantly advancing medical record curation and clinical workflows.

Chinese Mathematical Knowledge Entity Recognition Based on Linguistically Motivated Bidirectional Encoder Representation from Transformers

We assessed whether constructing a mathematical knowledge graph for a knowledge question-answering system or a course recommendation system, Named Entity Recognition (NER), is indispensable. The accuracy of its recognition directly affects the actual performance of these subsequent tasks. In order to improve the accuracy of mathematical knowledge entity recognition and provide effective support for subsequent functionalities, this paper adopts the latest pre-trained language model, LERT, combined with a Bidirectional Gated Recurrent Unit (BiGRU), Iterated Dilated Convolutional Neural Networks (IDCNNs), and Conditional Random Fields (CRFs), to construct the LERT-BiGRU-IDCNN-CRF model. First, LERT provides context-related word vectors, and then the BiGRU captures both long-distance and short-distance information, the IDCNN retrieves local information, and finally the CRF is decoded to output the corresponding labels. Experimental results show that the accuracy of this model when recognizing mathematical concepts and theorem entities is 97.22%, the recall score is 97.47%, and the F1 score is 97.34%. This model can accurately recognize the required entities, and, through comparison, this method outperforms the current state-of-the-art entity recognition models.

RAMIE: retrieval-augmented multi-task information extraction with large language models on dietary supplements.

To develop an advanced multi-task large language model (LLM) framework for extracting diverse types of information about dietary supplements (DSs) from clinical records. We focused on 4 core DS information extraction tasks: named entity recognition (2949 clinical sentences), relation extraction (4892 sentences), triple extraction (2949 sentences), and usage classification (2460 sentences). To address these tasks, we introduced the retrieval-augmented multi-task information extraction (RAMIE) framework, which incorporates: (1) instruction fine-tuning with task-specific prompts; (2) multi-task training of LLMs to enhance storage efficiency and reduce training costs; and (3) retrieval-augmented generation, which retrieves similar examples from the training set to improve task performance. We compared the performance of RAMIE to LLMs with instruction fine-tuning alone and conducted an ablation study to evaluate the individual contributions of multi-task learning and retrieval-augmented generation to overall performance improvements. Using the RAMIE framework, Llama2-13B achieved an F1 score of 87.39 on the named entity recognition task, reflecting a 3.51% improvement. It also excelled in the relation extraction task with an F1 score of 93.74, a 1.15% improvement. For the triple extraction task, Llama2-7B achieved an F1 score of 79.45, representing a significant 14.26% improvement. MedAlpaca-7B delivered the highest F1 score of 93.45 on the usage classification task, with a 0.94% improvement. The ablation study highlighted that while multi-task learning improved efficiency with a minor trade-off in performance, the inclusion of retrieval-augmented generation significantly enhanced overall accuracy across tasks. The RAMIE framework demonstrates substantial improvements in multi-task information extraction for DS-related data from clinical records.

Harnessing NLP to Investigate Biomarker Interactions and CVD Risks in Elderly Chronic Kidney Disease Patients.

Chronic kidney disease (CKD) significantly increases the risk of CVD diseases, particularly among elderly patients. Understanding the interaction between several biomarkers and cardiovascular (CVD) risks is crucial for improving patient outcomes and tailoring personalized treatment strategies. There is much more to learn about the intricate relationship between biomarkers and CVD risks in elderly CKD patients. Research aims to harness natural language processing (NLP) strategies to investigate the interaction between biomarkers and CVD risks in elderly patients with CKD. This research examined how changes in baseline values of four novel and classic cardiac biomarkers relate to the danger of CVD, and all-cause death in a large cohort of patients with CKD. Initially, medical data were collected from EHR of elderly CKD patients. NLP technique, such as Named Entity Recognition (NER), is used to extract the relevant biomarkers and CVD risk factors from the data. Statistical techniques were applied to examine the associations between biomarkers and CVD risks. The predictive models, using a combination of structured and NLP-extracted features demonstrated improved accuracy in forecasting CVD outcomes compared to traditional methods. This investigation highlights the critical role of specific biomarkers like PTH and FGF-23 in predicting CVD outcomes, providing insights into the possibility of using EHR data for better patient management and enhancing predictive models for this high-risk population.

An automated construction method of 3D knowledge graph based on multi-agent systems in virtual geographic scene

ABSTRACT Virtual geographic scenes can significantly facilitate the visualization and comprehension of the real world. However, the multitude of object types and complex relationships they contain make it challenging to display embedded geographic knowledge. Knowledge graphs help organize object entities and their relationships within scenes, which makes it easier to store and express geographic knowledge. Current methods for constructing knowledge graphs for virtual geographic scenes suffer from issues such as weak automation, low dimensionality, and poor cognitive efficiency. Thus, this paper presents a method for automatically constructing knowledge graphs by integrating intelligent systems. The method involves designing a multiagent system to extract entities and relationships hidden in virtual geographic scenes. In addition, a three-dimensional visualization approach is proposed for knowledge graphs constrained by spatiotemporal relationships. A virtual scene containing 2,183 objects was selected as the experimental area for constructing and displaying the knowledge graph. The results demonstrate that the method constructs a knowledge graph with 12,958 relationships within 300 s in common network and hardware settings. In addition, a knowledge graph offers better completeness at small scales compared to other geographic knowledge graphs, and it can also enhance the cognitive efficiency of understanding scene relationships.

Integration of Natural Language Processing (NLP) in MongoDB NoSQL: A New Era of Efficient Text Data Management

Advancements in computing power enable efficient management of large and complex datasets, leading to growing importance in big data analytics. This is evidenced in the progress demonstrated by word embedding’s in creating extensive index files of interconnected entities. Digital collections, which represent layered textual structures and multimodal communication annotations—ranging from linguistic to gestural data—point out the complexity of modern datasets. Evolving data forms call for new techniques for effective storage and management of data forms to be used in NLP applications. This paper analyzes six distinct Database Management Systems (DBMS) that will help in understanding optimum methods for dealing with the complex data types. Specifically, the paper looks into the areas of tokenization, semantic analysis, named entity recognition, sentiment analysis, text classification, and information retrieval. These were found to not be dominant in all tasks. We thus propose a web-based multi-database management system MDBMS that integrates specialized databases in various paradigms; this is scalable and adaptive. This MDBMS model integrates the strengths of different systems to meet the needs of heterogeneous datasets in NLP applications.

Risk Assessment of Typhoon Disaster Chain Based on Knowledge Graph and Bayesian Network

Typhoon disasters not only trigger secondary disasters, such as rainstorms and flooding, but also bring many negative impacts on the normal operation of urban infrastructure and the safety of people’s lives and property. In order to effectively prevent the risks of typhoon disaster chain, this paper proposes a joint entity and relation extraction model based on RoBERTa-Adv-GPLinker. Then, relying on the ontology theory and methodology, we establish a knowledge graph of typhoon disaster chain. The results show that the joint extraction model based on RoBERTa-Adv-GPLinker outperforms other baseline models in all assessment indexes. Moreover, the constructed knowledge graph of typhoon disaster chain includes secondary disasters and derived disaster impacts. This can largely depict the evolution process of typhoon disaster secondary derivations. On this basis, a risk assessment model of typhoon disaster chain based on Bayesian network is established. Taking Fujian Province as an example, the risk associated with the typhoon disaster chain is assessed, verifying the effectiveness of the method. This study provides a scientific basis for enhancing government emergency response capabilities and achieving sustainable regional development.

Building knowledge graph for relevant degree recommendations using semantic similarity search and named entity recognition

Career guidance is a critical and often daunting process, particularly during the transition from high school to higher education within the Moroccan education system. Faced with a vast array of university programs and career options, students frequently struggle to make informed decisions that align with their aspirations and skills. To address this challenge, our research introduces an innovative system that combines semantic similarity search with knowledge graph (KG) construction to enhance the precision and personalization of academic recommendations. By utilizing Sentence-BERT (SBERT) for semantic similarity, we generate embedding vectors that capture nuanced relationships between student profiles and degree descriptions. Subsequently, named entity recognition (NER) is applied to extract essential information such as skills, fields of study, and career opportunities from these profiles and descriptions. The extracted entities and their interrelationships are then structured into a coherent KG, stored in a Neo4j database, enabling efficient querying and visualization of complex data connections. This approach provides a transparent and explainable framework, ultimately delivering tailored advice that aligns with students’ individual needs and educational goals.

A Comparative Study of Deep Learning Approaches for Arabic Language Processing

Arabic is a difficult language for natural language processing (NLP) because of its complicated morphology, dialectal differences and the limited annotated resources. Although deep learning algorithms have reached state-of-the-art results in many NLP tasks, comprehensive comparative studies for Arabic remains scarce. This paper addresses this gap by systematically evaluating three prominent deep learning architectures - namely Recurrent Neural Networks (RNNs), Convolutional Neural Networks (CNNs) and Transformers - across five essential Arabic NLP tasks: i) mention of sentiment analysis, ii) named entity recognition, iii) machine translation, iv) text classification and v) dialect identification. We differ the performance of models trained from scratch with fine-tuned versions of AraBERT, a powerful Transformer-based model pre-trained on a large Arabic corpus. Our experiments employ the Arabic datasets already existing in literature and utilizes accuracy, F1-score and BLEU as the evaluation metrics. The results are indicative of the supremacy of Transformer-based models with regard to AraBERT that shows the highest scores in each task. Notably, AraBERT attains 95. 2% accuracy on sentiment analysis, which is higher than the accuracies of RNNs and CNNs. These improvements also become apparent in other tasks, with AraBERT ending up as the best among RNN, CNN and others. A 3-point difference for 3 BLEU in machine translation and 2. 3% F1-score on dialect recognition. This extensive assessment, in turn, highlights the advantages and disadvantages of deep learning architectures for Arabic NLP. The excellent AraBERT representation also demonstrates how transfer learning and synergy between Transformer architectures and large-scale pre-training can significantly help Arabic language technology development.

A joint entity and relation extraction framework for handling negative samples problems in named entity recognition

A joint entity and relation extraction framework for handling negative samples problems in named entity recognition

Enhanced Entity Recognition of Islamic Hadiths based-on Hybrid LSTM and AraBERT Model

This paper focuses on the training, evaluation and development of named entity recognition (NER) models designed for Islamic hadiths in Arabic Utilizing the Hadith Noor dataset, the study uses the BIO (Basic, In, Out) tagging scheme to classify words or tokens in NER tasks and the segmentation of the text into individual tokens. The right-skewed distribution revealed by examining the lengths of the Islamic hadiths revealed a right-skewed distribution, indicating that shorter texts are more common. Texts less than 100 words were most prevalent, followed by texts between 100 and 200 words, while texts longer than 200 words were rare. The dataset identifies eight types of entities, such as common names among narrators and locations. The study by training the three models AraBERT, LSTM and the hybrid model AraBERT-LSTM on Arabic text processing respectively, the hybrid model showed a performance, efficiency and accuracy of 0.981, outperforming the rest of the models, confirming its worth and reliability in NER tasks for natural language in Arabic, especially Islamic hadiths, which opens the way for exploring further investigations for future research in natural language processing.

Entity and relation extraction in the legal domain

Entity and relation extraction in the legal domain

How can language models assist with pharmaceuticals manufacturing deviations and investigations?

Large Language Models (LLM) such as the Generative-Pretrained-Transformer (GPT) and Large-Language-Model-Meta-AI (LLaMA) have attracted much attention. There is strong evidence that these models perform remarkably well in various natural language processing tasks. However, how to leverage them in domain-specific use cases and drive value remains an open question. Focusing on the digital transformation in the pharmaceutical manufacturing space, we propose that leveraging historical records of manufacturing deviations, as a mostly unstructured data source, in an organization can be beneficial for productivity, efficiency, quality, and compliance, specifically for addressing and closing new cases, or de-risking new manufacturing campaigns by identifying common themes and occurrences. Herein, by constructing an industrially relevant dataset, the ability of generative LLMs (e.g., GPT and Claude) and text embedding models in performing tasks related to pharmaceutical manufacturing deviations are studied. Generative models are evaluated for automating knowledge extraction from deviation reports in a mature organization, while embedding models are evaluated for identification of similar incidents from a large body of historical records using similarity analysis in vector space. Results show highly accurate outcomes for entity extraction tasks, especially with larger models, strong reasoning capabilities, as well as an interplay between the apparent reasoning and hallucination behavior of LLMs. Results also show the ability of embedding models for capturing semantics in certain deviation categories. Overall, these findings suggest significant potential for enhancing workflows in the pharmaceutical manufacturing through AI-driven tools, while also highlighting important questions that necessitate further research.

Named Entity Recognition: A Deep Dive

Generally speaking, Named Entity Recognition is one of the major tasks in NLP that involves the identification and classification of named entities mentioned in text, which include, among others, names of people, organizations, locations, dates, etc.

A large-scale Chinese patent dataset for information extraction

Information extraction is an important foundation for automated patent analysis. Deep learning methods show promising results for information extraction, the performance of such methods heavily depends on the available corpus. To promote research on Chinese information extraction and evaluate the performance of related systems, we present a novel dataset, named CPIE, and make it publicly available. The dataset consisting of five thousands records of Chinese patent documents. The data were annotated by a tagging team using an on-line annotation tools. The dataset was evaluated using a state-of-the-art information extraction method that involves named entity recognition and relationship classification. The results shed light on new challenges and promote information extraction research.

Tasaheel-v2: Development of Innovative Textual Analysis tool with Advanced Features

We introduce Tasaheel-v2, an automated tool specifically developed for Arabic Natural Language Processing (NLP)and textual analysis tasks. This work is an extension to the first version, Tasaheel-v1, comprised of traditional NLP tasksincluding stemming, segmentation, normalization, named entity recognition, and part of speech tagging. Furthermore, itincluded cutting-edge analytic methods, such as specific emotion, polarity, linguistics, and domain-specific word tagging. In thisnew innovative version, Tasaheel-v2, we introduce additional benefiting utilities designed to provide assistance for the Arabicresearch community. We specifically integrate another Arabic-specific POS tagger, a sentiment analyzer, and English-to-Arabictranslation functions. We leverage the utilities provided in Tasaheel to develop a machine-learning model designed to identifyArabic phishing emails and provide a thorough textual analysis to capture deceptive cues used to detect phishing linguisticpatterns. This tool contributes to the Arabic research domain by providing assistive NLP functions and textual analysis featuresall in one tool

The Use of Large Language Models in Combination with the Ontological Approach for the Synthesis of Natural Language Text

This research presented an approach based on the application of structured prompt instructions to large language models (LLM). Methodological foundations for using an ontology-driven system of structured prompts in interaction with ChatGPT were developed. The ChatGPT system allows expanding the knowledge base by obtaining new information from existing knowledge units based on a set of contexts. Thus, methods for generating meaningful responses in natural language were considered using a large language model using ontological approaches and natural language contexts. Using the proposed methodology, the OntoChatGPT system was developed, which effectively extracts entities from contexts, classifies them and generates appropriate responses. An experiment on the reverse synthesis of natural language sentences based on their ontological representation using large language models allows to clearly demonstrate the effectiveness of using the concept of large language models in dialogue systems. The study highlights the versatility of the methodology, emphasizing its applicability not only to ChatGPT but also to other chatbot systems based on LLMs, such as Google’s Bard utilizing the PaLM 2 LLM. The implementation of this technology is demonstrated using the Ukrainian language

Information extraction from green channel textual records on expressways using hybrid deep learning

The expressway green channel is an essential transportation policy for moving fresh agricultural products in China. In order to extract knowledge from various records, this study presents a cutting-edge approach to extract information from textual records of failure cases in the vertical field of expressway green channel. We proposed a hybrid approach based on BIO labeling, pre-trained model, deep learning and CRF to build a named entity recognition (NER) model with the optimal prediction performance. Eight entities are designed and proposed in the NER processing for the expressway green channel. three typical pre-trained natural language processing models are utilized and compared to recognize entities and obtain feature vectors, including bidirectional encoder representations from transformer (BERT), ALBERT, and RoBERTa. An ablation experiment is performed to analyze the influence of each factor on the proposed models. Used the survey data from the expressway green channel management system in Shaanxi Province of China, the experimental results show that the precision, recall, and F1-score of the RoBERTa-BiGRU-CRF model are 93.04%, 92.99%, and 92.99%, respectively. As the results, it is discovered that the text features extracted from pre-training substantially enhance the prediction accuracy of deep learning algorithms. Surprisingly, the RoBERTa model is highly effective in the task for the expressway green channel NER. This study provides a timely and necessary knowledge extraction on the Expressway Green Channel in terms of textual data, offering a systematical explanation of failure cases and valuable insights for future research.

The aluminum standard: using generative Artificial Intelligence tools to synthesize and annotate non-structured patient data

Background Medical narratives are fundamental to the correct identification of a patient’s health condition. This is not only because it describes the patient’s situation. It also contains relevant information about the patient’s context and health state evolution. Narratives are usually vague and cannot be categorized easily. On the other hand, once the patient’s situation is correctly identified based on a narrative, it is then possible to map the patient’s situation into precise classification schemas and ontologies that are machine-readable. To this end, language models can be trained to read and extract elements from these narratives. However, the main problem is the lack of data for model identification and model training in languages other than English. First, gold standard annotations are usually not available due to the high level of data protection for patient data. Second, gold standard annotations (if available) are difficult to access. Alternative available data, like MIMIC (Sci Data 3:1, 2016) is written in English and for specific patient conditions like intensive care. Thus, when model training is required for other types of patients, like oncology (and not intensive care), this could lead to bias. To facilitate clinical narrative model training, a method for creating high-quality synthetic narratives is needed.MethodWe devised workflows based on generative AI methods to synthesize narratives in the German language to avoid the disclosure of patient’s health data. Since we required highly realistic narratives, we generated prompts, written with high-quality medical terminology, asking for clinical narratives containing both a main and co-disease. The frequency of distribution of both the main and co-disease was extracted from the hospital’s structured data, such that the synthetic narratives reflect the disease distribution among the patient’s cohort. In order to validate the quality of the synthetic narratives, we annotated them to train a Named Entity Recognition (NER) algorithm. According to our assumptions, the validation of this system implies that the synthesized data used for its training are of acceptable quality.ResultWe report precision, recall and F1 score for the NER model while also considering metrics that take into account both exact and partial entity matches. Trained models are cautious, with a precision up to 0.8 for Entity Type match metric and a F1 score of 0.3.ConclusionDespite its inherent limitations, this technology has the potential to allow data interoperability by using encoded diseases across languages and regions without compromising data safety. Additionally, it facilitates the synthesis of unstructured patient data. In this way, the identification and training of models can be accelerated. We believe that this method may be able to generate discharge letters for any combination of main and co-diseases, which will significantly reduce the amount of time spent writing these letters by healthcare professionals.