Neural Network Methods for Natural Language Processing
... Additionally, advances in natural language processing (NLP) have enabled the automatic summarization and categorization of enormous amounts of data, greatly increasing efficiency and productivity. All of these developments highlight how important NLP is in influencing contemporary technological interactions and decision-making [1], [2]. Social media sites like Facebook, Instagram, Reddit, LinkedIn, and Twitter have become effective means of communication and are constantly producing vast amounts of user-generated material. ...
... Traditional natural language processing techniques frequently struggle with accuracy and efficacy due to the distinctive features of social media text. To better manage the subtleties and complexity present in social media information, machine learning (ML) and deep learning (DL) approaches have thus been used more and more [1], [2]. Text classification tasks on social media, such as sentiment analysis, topic classification, and spam detection, have long been handled by traditional machine learning techniques, such as Support Vector Machines (SVM), Naive Bayes classifiers, Decision Trees, Random Forests, and ensemble learning methods. ...
... When it comes to modeling sequential data and identifying contextual relationships in social media language, recurrent neural networks (RNNs), in particular Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRU), have shown remarkable efficacy [5]. These models are particularly well-suited for sentiment analysis, emotion detection, and disinformation identification tasks where context is crucial since they can comprehend contextual subtleties, emotional undertones, and sequential dependencies with ease [2]. By offering potent context-aware representations of textual data, transformer-based language models like BERT (Bidirectional Encoder Representations from Transformers), RoBERTa (Robustly optimized BERT pre-training approach), and GPT (Generative Pre-trained Transformer) have more recently transformed natural language processing. ...
Natural Language Processing (NLP) is an essential element of computational linguistics and artificial intelligence, enabling fluid interactions between humans and machines. Social networking networks produce substantial volumes of user-generated text daily, offering both opportunities and challenges for NLP researchers. Social media discourse's informal, dynamic, and context-dependent characteristics necessitate specific NLP techniques for precise processing and analysis. This study thoroughly examines NLP applications in social media, including essential tasks such as sentiment analysis, topic modeling, misinformation detection, and hate speech identification. It examines the influence of machine learning and deep learning methodologies, particularly transformer models, on the advancement of NLP capabilities. This study also emphasizes the ethical issues related to NLP-driven social media apps, including data privacy, algorithmic bias, and the regulation of misinformation. The paper continues by discussing emerging research paths, highlighting the necessity for adaptable and ethical NLP solutions in the changing social media environment.
... Neural networks (NNs) have gained significant attention over the past two decades due to their theoretical foundations and practical applications in various fields, including image processing [10], sound recognition [7], and natural language processing [14,15]. Ensuring the robustness and safety of NNs against adversarial perturbations to test inputs is a key challenge in safe machine learning. ...
... Consider an L-layer feed-forward NN [15] described as a function f : R n 0 → R n L+1 of form: ...
... x L ∈ R n 0 +N , appropriate coefficientsb ∈ R,q ∈ R n 0 +N , andQ ∈ S n 0 +N . As DeepSDP (3) is the dual problem of the SDP relaxation (16) of QCQP (15), the relationship between (15) and (3) is analogous to the relationship between the standard QCQP (4) and its dual SDP relaxation (6). Figure 1 shows the correspondence between (15) and (3). In the subsequent discussion, we refer to (16) as the primal SDP relaxation of (15), and distinguish it from the dual SDP relaxation, DeepSDP (3). ...
For verifying the safety of neural networks (NNs), Fazlyab et al. (2019) introduced a semidefinite programming (SDP) approach called DeepSDP. This formulation can be viewed as the dual of the SDP relaxation for a problem formulated as a quadratically constrained quadratic program (QCQP). While SDP relaxations of QCQPs generally provide approximate solutions with some gaps, this work focuses on tight SDP relaxations that provide exact solutions to the QCQP for single-layer NNs. Specifically, we analyze tightness conditions in three cases: (i) NNs with a single neuron, (ii) single-layer NNs with an ellipsoidal input set, and (iii) single-layer NNs with a rectangular input set. For NNs with a single neuron, we propose a condition that ensures the SDP admits a rank-1 solution to DeepSDP by transforming the QCQP into an equivalent two-stage problem leads to a solution collinear with a predetermined vector. For single-layer NNs with an ellipsoidal input set, the collinearity of solutions is proved via the Karush-Kuhn-Tucker condition in the two-stage problem. In case of single-layer NNs with a rectangular input set, we demonstrate that the tightness of DeepSDP can be reduced to the single-neuron NNs, case (i), if the weight matrix is a diagonal matrix.
... With rise in efficient computational resources, abstractive summarization also became more popular and implementable. Starting with vanilla feedforward neural network (FFNN), NLP gradually moved to recurrent neural network (RNN) and long-short term memory (LSTM), gated recurrent unit (GRU) networks to handle sequential data such as natural language text (Goldberg, 2017). Sequential data like natural language sentences have words with long-range dependencies, i.e., a word at the starting of a sequence may be important to the last or later word in a sentence. ...
... Distributional hypothesis is the idea that words that occur in related contexts tend to have the same meaning (Firth, 1957;Harris, 1954). This hypothesis paved way for distributed representation of words with the aim to make machines better understand the context of a word, understanding similar words based on their meaning, and differentiating between them based on their contexts and semantics (Goldberg, 2017). Algorithms for distributed representations learn word representations from large scale monolingual corpora through unsupervised learning. ...
... SL also has a variant in weak or distant supervised learning which involves a semi-automatic process of labeling unlabeled data through additional sources like databases and dictionaries (Hedderich et al., 2020(Hedderich et al., , p. 2548. Most of these SL and USL methods are used widely in natural language classification tasks (Goldberg, 2017). ...
This thesis presents Abstractive Text Summarization models for contemporary Sanskrit prose. The first chapter, titled Introduction, presents the motivation behind this work, the research questions, and the conceptual framework. Sanskrit is a low-resource inflectional language. The key research question that this thesis investigates is what the challenges in developing an abstractive TS for Sanskrit. To answer the key research questions, sub-questions based on four different themes have been posed in this work. The second chapter, Literature Review, surveys the previous works done. The third chapter, data preparation, answers the remaining three questions from the third theme. It reports the data collection and preprocessing challenges for both language model and summarization model trainings. The fourth chapter reports the training and inference of models and the results obtained therein. This research has initiated a pipeline for Sanskrit abstractive text summarization and has reported the challenges faced at every stage of the development. The research questions based on every theme have been answered to answer the key research question.
... Recently, deep learning has emerged as a promising approach that can automatically learn representations from data (Goldberg, 2022). Moreover, deep learning techniques like CNNs (Yao et al., 2020;Renjith et al., 2022;Tadesse et al., 2019), LSTMs (Haque et al., 2022;Tadesse et al., 2019;Renjith et al., 2022;Ji et al., 2018;Ma et al., 2018), BiLSTM (Haque et al., 2022;Zhang et al., 2022;He and Lin, 2016) and DLSTMAttention (Zhang et al., 2022;Renjith et al., 2022) have been applied in detecting suicidal ideation, with competitive performance. ...
... The recent advances in neural network-based language models have demonstrated substantial improvements across a wide range of natural language processing tasks (Goldberg, 2022). In particular, the introduction of Transformer architectures (Vaswani et al., 2017) led to unprecedented progress in semantic and syntactic modeling capabilities. ...
Suicidal ideation is a serious health problem affecting millions of people worldwide. Social networks provide information about these mental health problems through users' emotional expressions. We propose a multilingual model leveraging transformer architectures like mBERT, XML-R, and mT5 to detect suicidal text across posts in six languages - Spanish, English, German, Catalan, Portuguese and Italian. A Spanish suicide ideation tweet dataset was translated into five other languages using SeamlessM4T. Each model was fine-tuned on this multilingual data and evaluated across classification metrics. Results showed mT5 achieving the best performance overall with F1 scores above 85%, highlighting capabilities for cross-lingual transfer learning. The English and Spanish translations also displayed high quality based on perplexity. Our exploration underscores the importance of considering linguistic diversity in developing automated multilingual tools to identify suicidal risk. Limitations exist around semantic fidelity in translations and ethical implications which provide guidance for future human-in-the-loop evaluations.
... The foundation of automated content generation was laid by early language models such as Word2Vec and GloVe, which represented words as continuous vectors based on their semantic relations in large corpora [9,10]. These models enabled basic sentencelevel content generation and served as the backbone for tasks such as sentiment analysis and machine translation [19]. Sequence-to-sequence (Seq2Seq) models, such as the one introduced by [20], further pushed the envelope by allowing end-to-end neural networks to translate entire sequences of words into another sequence, expanding possibilities for more complex text generation. ...
... Obtain prediction probabilities for "suitable" and "unsuitable" 13: Set threshold for determining suitability (e.g., 0.5 probability) 14: if prediction > threshold for "suitable" then 15: Append "Suitable" to sentence_classification_results 16: else 17: Append "Unsuitable" to sentence_classification_results 18: end if 19 6: for each sentence and its corresponding classification result in generated_story do 7: if classification_result == "Unsuitable" then 8: Pass the unsuitable sentence to the LLM for reframing 9: Provide context about child-appropriate content in the prompt 10: Example prompt: "Reframe the following sentence to make it suitable for children: <unsuitable_sentence>" 11: Capture the LLM's reframed sentence as modified_sentence 12: Append modified_sentence to reframed_story 13: else 14: Append the original sentence to reframed_story 15: end if 16: end for 17: Combine reframed_story sentences back into a full story 18: Feed reframed_story into the BERT Classifier for classification 19: Initialize suitable_story as False 20: while suitable_story == False do 21: Run the BERT Classifier on reframed_story to classify sentences 22: if all sentences are classified as "Suitable" then 23: Set suitable_story = True 24: else 25: Repeat steps 4 to 7 for any new unsuitable sentences 26: end if 27: end while 28: Return reframed_story (where all sentences are classified as suitable) 29: End ...
In today’s digital age, ensuring the appropriateness of content for children is crucial for their cognitive and emotional development. The rise of automated text generation technologies, such as Large Language Models like LLaMA, Mistral, and Zephyr, has created a pressing need for effective tools to filter and classify suitable content. However, the existing methods often fail to effectively address the intricate details and unique characteristics of children’s literature. This study aims to bridge this gap by developing a robust framework that utilizes fine-tuned language models, classification techniques, and contextual story generation to generate and classify children’s stories based on their suitability. Employing a combination of fine-tuning techniques on models such as LLaMA, Mistral, and Zephyr, alongside a BERT-based classifier, we evaluated the generated stories against established metrics like ROUGE, METEOR, and BERT Scores. The fine-tuned Mistral-7B model achieved a ROUGE-1 score of 0.4785, significantly higher than the base model’s 0.3185, while Zephyr-7B-Beta achieved a METEOR score of 0.4154 compared to its base counterpart’s score of 0.3602. The results indicated that the fine-tuned models outperformed base models, generating content more aligned with human standards. Moreover, the BERT Classifier exhibited high precision (0.95) and recall (0.97) for identifying unsuitable content, further enhancing the reliability of content classification. These findings highlight the potential of advanced language models in generating age-appropriate stories and enhancing content moderation strategies. This research has broader implications for educational technology, content curation, and parental control systems, offering a scalable approach to ensuring children’s exposure to safe and enriching narratives.
... Natural Language Processing is a field of artificial intelligence focusing on the interaction between computers and human language. Its primary purpose is to enable machines to understand, interpret and generate human language naturally and effectively (Goldberg, 2022;Rayhan, 2024;Speer et al., 2024). It uses algorithms and statistical models to analyse and process large amounts of linguistic data. ...
... El Natural Language Processing es un campo de la inteligencia artificial que se enfoca en la interacción entre los ordenadores y el lenguaje humano. Su objetivo principal es permitir que las máquinas entiendan, interpreten y generen lenguaje humano de manera natural y efectiva (Goldberg, 2022;Rayhan, 2024;Speer et al., 2024). Utiliza algoritmos y modelos estadísticos para analizar y procesar grandes cantidades de datos lingüísticos. ...
A spanish version of the article is provided (see section before Acknowledgements) As scientific output grows, systematic reviews have become essential yet increasingly challenging. Our approach to this protocol aims to make this process more effective, efficient and accessible to researchers worldwide, including those in developing countries. We developed a tool to complement human judgment in the screening phase using pre-trained language models and natural language processing techniques. This tool generates text embeddings and calculates semantic similarities, prioritizing potentially relevant articles. The goal is to utilise the similarity ranking instead of reviewing articles randomly or following the relevance sort option of search engines like WOS or Scopus. Coders can start with those closest to the category/categories of interest and progressively move towards the more distant ones. This approach would save time and effort while reducing the fatigue and biases of the coders. The models we have tested in this research are all-MiniLM-L6-v2, all-distilroberta-v1, all-mpnet-base-v2, paraphrase-multilingual-mpnet-base-v2, distiluse-base-multilingual-cased-v1, all-MiniLM-L12-v2, allenai-specter, allenai/scibert_scivocab_uncased, distilbert-base-nli-mean-tokens, roberta-base-nli-stsb-mean-tokens, distiluse-base-multilingual-cased-v2, paraphrase-multilingual-MiniLM-L12-v2, stsb-roberta-large, bert-base-nli-mean-tokens. The method was implemented using limited computational resources and open-source software, ensuring accessibility for research teams with restricted economic resources. Results indicate a possible reduction in screening time and improved consistency in article selection. The tool demonstrated utility in classifying relevant studies and would facilitate more comprehensive reviews. By providing a low-cost solution, we aim to level the playing field in global research, enabling researchers from diverse economic backgrounds to participate more fully in producing scientific knowledge.
... As an important branch of artificial intelligence, it also occupies an increasingly important position in the field of data processing and is nowadays known and applied by most people. Natural Language Processing is mainly divided into two processes: Natural Language Understanding (NLU) and Natural Language Generation (NLG) [4][5][6]. NLU mainly focuses on understanding the meaning of the text. Specifically, each word and structure needs to be understood [7]. ...
... In this chapter, the encoding layer is denoted as s E , the target word is denoted as t , the context in which it is located is denoted as s , the information about the start position where the target word is located is noted as st , the information about the end position is noted as en , and the last layer of the BERT is denoted as t H . The resulting encoding of the target word t r can be denoted as Equation (6). Combined with the position P of the syntactic or semantic role information obtained by the extraction layer, the i th block of syntactic or semantic role information i sk r associated with the target word is denoted as Equation (7): ...
Semantic analysis, a crucial aspect of natural language processing, encounters numerous practical challenges due to the limitations of its current technology. Therefore, this paper enhances the traditional semantic analysis technology by developing a frame recognition model that integrates syntactic and semantic roles, a text semantic feature extraction model, and an audio/video information extraction model with a multimodal inter-modal cross-attention mechanism. These models are then integrated to jointly construct an improved model for semantic analysis, which is based on deep neural networks. The paper examines the model’s improvement effect in semantic role labeling, text classification, and information extraction. The F1 values of this paper’s model on the Wall Street Journal and Brown test sets are 90.4% and 81.4%, respectively, which are the highest semantic role recognition annotation accuracy rates. The HL, P, R, and F1 values of this paper’s model on the three datasets, on the other hand, are the best results among all models, and it has the best text categorization effect. This paper’s model has a 95.3% accuracy rate in detecting theme subtitles. The recognition accuracy of simple and complex backgrounds is 95.7% and 94.1%, respectively. After the information extraction method of this paper’s model underwent error correction, the accuracy of ASR recognition increased by 18.55%.
... Bunlardan biri metin sınıflandırmadır. Örneğin, e-posta servisleri, gelen iletileri spam veya normal olarak sınıflandırmak için doğal dil işleme algoritmalarını kullanır (Bird ve ark., 2009;Goldberg, 2022). Doğal dil işleme alanında bir üretken yapay zeka teknolojisi olan ChatGPT, bu alandaki öncü bir yenilik olarak önemli bir ilerlemeyi temsil etmektedir. ...
Amaç: Bu çalışma, spor bilimlerinde yapay zeka kullanımının teorik çerçevenin ötesine geçerek uygulamalara nasıl yansıdığını ve spor sektöründeki kullanımını incelemeyi amaçlamaktadır. Yapay zeka destekli yenilikçi çözümlerin etkinliği, spor bilimleri alanındaki mevcut uygulamalar çerçevesinde değerlendirilecektir
Yöntem: Bu araştırmada nitel yöntemlerden doküman analizi ve örnek olay deseni bir arada kullanılmıştır. Doküman analizinde, yapay zeka kavramı ve bileşenleri literatür taraması yoluyla incelenmiştir. Örnek olay deseni kapsamında ise, spor sektöründe faaliyet gösteren kuruluşlar, federasyonlar, organizasyonlar, kulüpler, geliştirici şirketler ve analiz firmalarının yapay zeka uygulamaları ele alınmıştır.
Bulgular: Spor bilimleri alanında kullanılan ve geliştirilmiş olan 18 yapay zeka uygulaması şekiller şablonunda belirtilmiş ve kullanım amaçları ile birlikte açıklanmıştır.
Sonuç: Sonuç olarak, yapay zeka uygulamalarının spor endüstrisine ve spor bilimleri alanına entegrasyonu hızla artmaktadır. Bu teknolojilerin kullanımı, teorik bilgiyi pratik uygulamalara dönüştürerek spor ekosisteminde performans analizi, sakatlık önleme ve antrenman planlaması gibi alanlarda önemli katkılar sağlamaktadır. Önümüzdeki yıllarda yapay zekanın spor bilimlerindeki rolü daha da güçlenecek ve alanın geleceğini belirleyen temel faktörlerden biri olacaktır.
... Given a set of L input variables X = (X 1 , . . . , X L ) that can be numerical or categorical, or a combination of them, and the corresponding output Y , we have to focus on the hyperparameters of the NNs, that is those settings that have to be set before the parameters are learnt in the training process, see Goldberg (2017) and Prince (2023). These hyperparameters are ...
In recent decades, analysing the progression of mortality rates has become very important for both public and private pension schemes, as well as for the life insurance branch of insurance companies. Traditionally, the tools used in this field were based on stochastic and deterministic approaches that allow extrapolating mortality rates beyond the last year of observation. More recently, new techniques based on machine learning have been introduced as alternatives to traditional models, giving practitioners new opportunities. Among these, neural networks (NNs) play an important role due to their computation power and flexibility to treat the data without any probabilistic assumption. In this paper, we apply multi-task NNs, whose approach is based on leveraging useful information contained in multiple related tasks to help improve the generalized performance of all the tasks, to forecast mortality rates. Finally, we compare the performance of multi-task NNs to that of existing single-task NNs and traditional stochastic models on mortality data from 17 different countries.
... So far, artificial neural networks are part of machine learning models, which are frequently used in the domain of Natural Language Processing (Goldberg, 2017). In finance, one of the most prevalent artificial neural networks is the multilayer perceptron (MLP) model (Kumar & Ravi, 2016). ...
This study develops models that predict banks’ stock price crash risk using novel machine learning techniques. A key element of our approach is that we retrieve textual information from ECB presidents’ speeches. To this end, we employ quarter-bank level data and various measures for stock price crash risk, ensuring the robustness of our findings. First, we find that the machine learning models can generally perform better than the simple regressions. Next, our results also suggest that textual information from the ECB president’s speeches has significant predictive power. Finally, when we jointly use textual information and macro-financial variables as inputs, the performance of our models is substantially increased compared to models using a single type of input. Our empirical findings provide significant policy implications for investors and policymakers as they can help regulators assess the financial system’s stability and identify any potential systemic risks, allowing them to take proactive measures to prevent or mitigate a financial crisis.
... With the trend of world integration, communication between different languages and cultures, as well as the international academic circle, is getting closer and closer [9]. In this process, people will inevitably come into contact with a large amount of information in non-native languages, and when information seekers are faced with this kind of information, the rudiments of non-native languages and the intricacies of network information bring great inconvenience to them, and it is difficult for them to get the useful information they need quickly because reading through the huge amount of pages of information will definitely consume a lot of time and energy [10][11][12]. ...
British and American Imagist poetry has a profound influence on the entire Western literary world, and it is the poetry form of the British anti-Romantic literary movement. The language expression of the poetry is more implicit, and the language structure is concise and clear, with very strong modernist characteristics. In this paper, we design a strategy for translating English and American poetry imagery with the assistance of natural language processing technology, i.e., we construct a machine translation model based on Transformer’s chapter context validity recognition through the corpus to realize the accurate translation of English and American poetry imagery, and experimentally analyze the effect of the model. The method in this paper achieves the expected performance, with a maximum improvement of +1.99 BLEU compared to the sentence-level baseline model and a maximum improvement of +0.94 BLEU compared to the chapter-level baseline model, and achieves the optimum among a series of typical chapter-level translation models compared. Through the statistics of the deep meaning of imagery, it is known that the same deep meaning can be expressed by things of different meaning categories, and through the ratio of imagery type and frequency, it is known that poets will choose different imagery to express the same deep meaning. At the same time, in English and American poetry, a large number of rhetorical devices, such as borrowing and simile, are used. The deeper meanings of the imagery mostly reflect negative and painful emotions. The deeper meanings of many of the imagery are extremely rich, which reflects the polysemous nature of poetry. This paper lays the foundation for a better study of a chapter translation model that can fully perceive and efficiently utilize chapter context.
... Neural NLG is usually considered to be a sequence-to-sequence task, where an input sequence (input texts for summarisation, time series for data-to-text) is converted into an output sequence of words (ie, a text). Early neural models for NLG [72] used recurrent neural networks (RNNs), which are a type of neural network which iterates through arbitrary-length sequences of tokens (words). Long short-term memory (LSTM) architectures modify the core RNN to give the network 'memory' which makes it easier for the network to consider already-output words when it decides on the next word to output. ...
This book provides a broad overview of Natural Language Generation (NLG), including technology, user requirements, evaluation, and real-world applications. The focus is on concepts and insights which hopefully will remain relevant for many years, not on the latest LLM innovations. It draws on decades of work by the author and others on NLG. The book has the following chapters: Introduction to NLG; Rule-Based NLG; Machine Learning and Neural NLG; Requirements; Evaluation; Safety, Maintenance, and Testing; and Applications. All chapters include examples and anecdotes from the author's personal experiences, and end with a Further Reading section. The book should be especially useful to people working on applied NLG, including NLG researchers, people in other fields who want to use NLG, and commercial developers. It will not however be useful to people who want to understand the latest LLM technology. There is a companion site with more information at https://ehudreiter.com/book/
... As NLP systems continue to improve, the collaboration between computer science and linguistics grows more critical, as it bridges the gap between human language and machine understanding. The study of NLP requires not only a deep understanding of how language works but also the computational power to handle vast amounts of data and complex models, making it an interdisciplinary field at the forefront of AI research [5,6]. to extract from a text. ...
Tacit knowledge, often implicit and deeply embedded within individuals and organizational practices, is critical for fostering innovation and decision-making in knowledge management systems (KMS). Converting tacit knowledge into explicit forms enhances organizational effectiveness by making this knowledge accessible and reusable. This paper presents a comparative analysis of natural language processing (NLP) algorithms used for document and report mining to facilitate tacit knowledge conversion. This study focuses on algorithms that extract insights from semi-structured and document-based natural language representations, commonly found in organizational knowledge artifacts. Key NLP strategies, including text mining, information extraction, sentiment analysis, clustering, classification, recommendation systems, and affective computing, are evaluated for their effectiveness in identifying and externalizing tacit knowledge. The findings highlight the relative strengths and limitations of these techniques, offering practical guidance for selecting suitable algorithms based on organizational needs. Additionally, this paper identifies challenges and emerging opportunities for advancing NLP-driven tacit knowledge conversion, providing actionable insights for researchers and practitioners aiming to enhance KMS capabilities.
... Numerical values are encoded or replaced with placeholders to maintain the semantic integrity of the text [61]. Outof-vocabulary words are managed through tokenisation or character-level representations [62], while padding and truncation ensure uniform sequence lengths, which is crucial for text classification [63]. Pretrained word embeddings, such as Word2Vec [64], can be used to initialise the embedding layers of deep learning models or be fine-tuned during training. ...
Emotion recognition and generation have emerged as crucial topics in Artificial Intelligence research, playing a significant role in enhancing human-computer interaction within healthcare, customer service, and other fields. Although several reviews have been conducted on emotion recognition and generation as separate entities, many of these works are either fragmented or limited to specific methodologies, lacking a comprehensive overview of recent developments and trends across different modalities. In this survey, we provide a holistic review aimed at researchers beginning their exploration in emotion recognition and generation. We introduce the fundamental principles underlying emotion recognition and generation across facial, vocal, and textual modalities. This work categorises recent state-of-the-art research into distinct technical approaches and explains the theoretical foundations and motivations behind these methodologies, offering a clearer understanding of their application. Moreover, we discuss evaluation metrics, comparative analyses, and current limitations, shedding light on the challenges faced by researchers in the field. Finally, we propose future research directions to address these challenges and encourage further exploration into developing robust, effective, and ethically responsible emotion recognition and generation systems.
... Natural Language Processing (NLP) has been extensively studied in various contexts and applications [34,54,70,73,74,94,137,215,231,232,348,355]. ...
A hypergraph extends this idea by allowing edges, referred to as hyperedges, to connect any number of vertices [30]. This paper explores superhypergraphs, an extension of hypergraphs incorporating superedges and supervertices. For example, Arboreal Superhypergraphs, Molecular superhypergraphs, and Probabilistic SuperHyperGraphs illustrate diverse structural types that can be modeled using superhypergraphs. We introduce the Generalized n-th Powerset, a formalized framework enabling broader mathematical applications while preserving the traditional n-th powerset structure. And we provide a brief exploration of Natural Hyperlanguage Processing, an extended framework of Natural Language Processing that leverages the concept of hyperlanguage for advanced applications. By extending hypergraph concepts to superhypergraphs, this work aims to advance their study and practical applicability.
... Natural Language Processing (NLP) involves enabling computers to understand, interpret, and generate human language for purposes of communication and analysis [25,50]. NLP has been extensively studied in various contexts and applications [25,33,50,52,53,75,108,175,185,187,199,321,331]. ...
... Natural Language Processing (NLP) involves enabling computers to understand, interpret, and generate human language for purposes of communication and analysis [25,50]. NLP has been extensively studied in various contexts and applications [25,33,50,52,53,75,108,175,185,187,199,321,331]. ...
This paper explores the integration of uncertainty frameworks such as Fuzzy, Neutrosophic, and Plithogenic sets into Large Language Models (LLMs) and Natural Language Processing (NLP). We propose novel models, including Large Uncertain Language Models and Natural Uncertain Language Processing, to enhance linguistic representations and processing capabilities. Furthermore, we extend the theoretical foundation of LLMs and NLP by incorporating Hyperstructures and Superhyperstructures, enabling higher-order generalizations and hierarchical modeling. These advancements provide new perspectives for addressing uncertainty and complexity in language understanding and processing. While the paper focuses on theoretical generalizations, practical validation through computational experiments remains an important direction for future work.
... and Natural Languages[65,98,257], have been extensively explored in the literature. With advancements in fields such as machine learning, these studies have garnered significant attention, particularly in areas like Natural Language Processing (cf.[65,98,101,168,241,257,258,301,421,432,438,444]).Technologies such as ChatGPT also incorporate these concepts in their applications ([7,55,133,189,212,248]). Moreover, related ideas like HyperLanguage[69,70,132,135,144] and SuperHyperLanguage[135,144] are well-documented and serve as foundational concepts for further exploration. ...
This work investigates the evolution of traditional set theory to address complex and ambiguous real-world phenomena. It introduces hierarchical hyperstructures and superhyperstructures, where superhyper-structures are formed by iteratively applying power sets to create nested abstractions. The focus is placed on three foundational set-based frameworks-Fuzzy Sets, Neutrosophic Sets, and Plithogenic Sets-and their extensions into Hyperfuzzy Sets, HyperNeutrosophic Sets, and Hyperplithogenic Sets. These extensions are applied to various domains, including Statistics, TOPSIS, K-means Clustering, Evolutionary Theory, Topological Spaces, Decision Making, Probability, and Language Theory. By exploring these generalized forms, this paper seeks to guide and inspire further research and development in this rapidly expanding field.
... El modelo LDA se apoya en representaciones de las preguntas tipo test basadas en un modelo bolsa de palabras (del inglés bag-of-words) Goldberg, 2022. Siguiendo este modelo, cada una de las preguntas a analizar se corresponde con un vector de caracter´ısticas numéricas, atendiendo al siguiente procedimiento de vectorización: ...
Este estudio analiza el uso del modelo generativo GPT-4 para generar preguntas tipotest en asignaturas de grado, mediante la lectura de material en PDF y otros recursoseducativos que se distribuyen al alumnado en clase. Para diversas asignaturas, se generaronpreguntas tipo test de forma autom´atica y se evalu´o de forma cuantitativa ycualitativa la idoneidad de las mismas. Los resultados muestran que GPT-4 ofrece unaforma r´apida y flexible de generar cuestionarios autoevaluables y ex´amenes, lo cualagiliza el proceso de evaluaci´on y el aprendizaje de los estudiantes. Sin embargo, lapertinencia del contenido generado depende en gran medida de la informaci´on presenteen la red, as´ı como de los materiales con los que el docente alimenta al modelo, siendoen todo caso necesaria la supervisi´on humana
... Natural Language Processing (NLP) has been extensively studied in various contexts and applications [28,44,58,61,62,80,118,188,203,204,300,307]. ...
A hypergraph extends this idea by allowing edges, referred to as hyperedges, to connect any number of vertices [25]. This paper explores superhypergraphs, an extension of hypergraphs incorporating superedges and supervertices. For example, Arboreal Superhypergraphs, Molecular superhypergraphs, and Probabilistic SuperHyperGraphs illustrate diverse structural types that can be modeled using superhypergraphs. We introduce the Generalized n-th Powerset, a formalized framework enabling broader mathematical applications while preserving the traditional n-th powerset structure. By extending hypergraph concepts to superhypergraphs, this work aims to advance their study and practical applicability.
... Incorporating State-of-the-Art Techniques: It is critical to stay up to date on the newest advances in RL in order to achieve optimum efficiency in RL-NLG networks. This may entail utilizing cutting-edge algorithms such as Soft Actor-Critic (SAC), Dual Delayed DDPG as (TD3), or other advanced methods [20]. The goal is to use the most effective strategies for the NLG scenario at hand. ...
Abstract. Finally, utilizing an interpretative framework as well as a deductive method this study studied the use of reinforcement learning strategies in the development of natural languages. Although secondary data collection provided a solid foundation, the descriptive method allowed for a complete study. The technical solutions used included the incorporation of RL computer programs, advanced text
processing, linguistic analysis, contextual adaptability, and moral concerns.As the
consequence of our research, we now have a better understanding of RL-NLG systems,
which opens the way to more sophisticated and environmentally conscious
text production. This research not only enhances the field of NLG, although it also
emphasizes the importance of ethical as well as objective applications for artificial
intelligence. The insights have application that encompass intelligent machines to
the generation of customized content.
... In our study, gamma is defined as 0.1; exception is the case where we use only financial variables where gamma is equal to 1. 6 4.2. Multilayer perceptron Artificial neural networks are a widely-used category of machine learning models especially in the domain of Natural Language Processing (Goldberg, 2017). In finance, one of the most popular artificial neural networks is the multilayer perceptron (MLP) model (Kumar and Ravi, 2016). ...
We combine machine learning algorithms (ML) with textual analysis techniques to forecast bank stock returns. Our textual features are derived from press releases of the Federal Open Market Committee (FOMC). We show that ML models produce more accurate out-of-sample predictions than OLS regressions, and that textual features can be more informative inputs than traditional financial variables. However, we achieve the highest predictive accuracy by training ML models on a combination of both financial variables and textual data. Importantly, portfolios constructed using the predictions of our best performing ML model consistently outperform their benchmarks. Our findings add to the scarce literature on bank return predictability and have important implications for investors. JEL classification: C63, E58, G17, G21, G40
... Also, the authors [19] mention "sentiment annotation" (which is to label emotion, opinion, and sentiment inherent within a text) as one of the most trending annotations in NLP or text analysis that can describe (textual) data into positive sentiment, neutral sentiment, or negative sentiment [1], [4], [5], [6], [8]. Whilst BoW [34] is concerned with known words in a textual document, the IDF of a word denotes how common or sporadic a word is in a corpus [19]. The multiplication of the TF and IDF features (also known as TF-IDF) [35] measures/estimates the impact or significance of each word (term) in a document based on their weights or valence as illustrated in this study. ...
Emotions classification or valence extraction in textual datasets, e.g. the students' opinion data, is becoming an emerging topic aimed at understanding the impact or intensities of words (terms) used by the users in the different contexts/applications. The method (emotional valence) for textual data scrutiny and exploration have proved to be central to the human experience analysis. To this effect, this study implements a text mining approach that determines the impact of the emotional valence (textual data quantification) shown by the students in their feedback provided during the semester course (pre and post) to determine its (emotional classification) relation or interconnectedness with the students' learning outcome and performance. The proposed method is designed based on the appraisal theories and component process model (CPM) that studies the degree of pleasantness or goal achievement as an effect of valence judgements. Technically, the study identifies the top terms in the students' data that can be used to draw insights or understand the learning experiences or performance by using Corpus feature selection and Term document matrix word processing libraries in R programming software. Also, it utilized the emotional valence score (quantified data) extracted from the analyzed (textual) data to statistically investigate the correlation and effect of the emotions scores (polarization or intensity of words) expressed by the students with their final grade or learning outcome using Sentiment Analysis package and Statistical analysis methods such as the Spearman's rho (ρ), Kendall's tau (), and Kruskal-Wallis H-test in R. The results shows that while in overall the emotional valence of the students do not influence or determine the outcome of their study or final average grade (p>0.05). When considering the analyzed comments: Pre-and Post-Course; it found that emotional scores (valence) expressed by the students in the Post-Course are more closely related or linked to the final grades than the Pre-Course comments. The paper empirically sheds light on both the pedagogical and socio-technical implications of the findings and result toward the achievement of higher levels of students learning outcomes or performance and a sustainable educational practice.
Keywords— text mining, emotional valence, student evaluation, textual data, educational data, higher education, educational innovation
... N-gram language models are commonly utilized for meeting latency demands owing to their ease of implementation and explainability [31,52]. With the advancements in deep learning technologies, models based on deep learning have been progressively incorporated into language modeling [9,22,68]. However, there has been limited research on contextual language modeling for text entry systems. ...
Text entry is a critical capability for any modern computing experience, with lightweight augmented reality (AR) glasses being no exception. Designed for all-day wearability, a limitation of lightweight AR glass is the restriction to the inclusion of multiple cameras for extensive field of view in hand tracking. This constraint underscores the need for an additional input device. We propose a system to address this gap: a ring-based mid-air gesture typing technique, RingGesture, utilizing electrodes to mark the start and end of gesture trajectories and inertial measurement units (IMU) sensors for hand tracking. This method offers an intuitive experience similar to raycast-based mid-air gesture typing found in VR headsets, allowing for a seamless translation of hand movements into cursor navigation. To enhance both accuracy and input speed, we propose a novel deep-learning word prediction framework, Score Fusion, comprised of three key components: a) a word-gesture decoding model, b) a spatial spelling correction model, and c) a lightweight contextual language model. In contrast, this framework fuses the scores from the three models to predict the most likely words with higher precision. We conduct comparative and longitudinal studies to demonstrate two key findings: firstly, the overall effectiveness of RingGesture, which achieves an average text entry speed of 27.3 words per minute (WPM) and a peak performance of 47.9 WPM. Secondly, we highlight the superior performance of the Score Fusion framework, which offers a 28.2% improvement in uncorrected Character Error Rate over a conventional word prediction framework, Naive Correction, leading to a 55.2% improvement in text entry speed for RingGesture. Additionally, RingGesture received a System Usability Score of 83 signifying its excellent usability.
This Chapter presents a comprehensive examination of Natural Language Processing [NLP] techniques for sentiment analysis in social media contexts, addressing the unique challenges and opportunities presented by user-generated content on social platforms. We trace the evolution of sentiment analysis from early lexicon-based approaches through modern transformer architectures, highlighting the technological advancements that have revolutionized our ability to understand and analyze social media sentiment. The Chapter provides detailed insights into advanced NLP techniques, preprocessing methodologies, and architectural considerations for implementing robust sentiment analysis systems. We also explore evaluation frameworks and metrics for assessing system performance, along with crucial implementation considerations for real-world applications. The Chapter concludes with an examination of emerging trends and future directions, including privacy-preserving techniques and ethical considerations in sentiment analysis.
Online exams have become a crucial part of modern education and recruitment, offering convenience and flexibility for both students and organizations. However, they face significant challenges such as cheating, impersonation, and unauthorized access to resources, which undermine the integrity and fairness of the process [1][3]. These issues create a need for reliable solutions to maintain the authenticity of online assessments. The Online Exam Proctoring system ensures a secure and fair testing environment through real-time monitoring, detecting suspicious activities like tab-switching, sound anomalies, and facial recognition-based alerts to prevent cheating [2][4][7]. With role-based access control (RBAC), examiners can efficiently manage exams and sensitive data, while students take exams in a proctored interface [10]. By automating monitoring and analysis, the system reduces human effort and errors [6][8], offering a reliable and transparent solution for online assessments and ensuring the integrity of the process for educational institutions and organizations [1][4].
This project introduces an AI-powered Banking Bot that automates banking tasks and improves customer support. It uses Artificial Intelligence (AI) and Machine Learning (ML) to handle important functions like money transfers, balance inquiries, account creation, and PIN generation. By automating these processes, the system ensures faster, more accurate, and reliable banking with minimal human involvement while maintaining strong security measures for transactions. Beyond automation, the bot also acts as a virtual assistant, helping customers navigate banking services, answer questions, and resolve common issues. With Natural Language Processing (NLP), it understands user queries naturally, making interactions smooth and user-friendly. By combining automation and real-time support, the Banking Bot offers a seamless 24/7 banking experience, meeting customer needs while improving efficiency. This technology transforms traditional banking, making it more secure, accessible, and personalized.
The rapid advancement of artificial intelligence (AI) technology is profoundly transforming the information ecosystem, reshaping the ways in which information is produced, distributed, and consumed. This study explores the impact of AI on the information environment, examining the challenges and opportunities for sustainable development in the age of AI. The research is motivated by the need to address the growing concerns about the reliability and sustainability of the information ecosystem in the face of AI-driven changes. Through a comprehensive analysis of the current AI landscape, including a review of existing literature and case studies, the study diagnoses the social implications of AI-driven changes in information ecosystems. The findings reveal a complex interplay between technological innovation and social responsibility, highlighting the need for collaborative governance strategies to navigate the tensions between the benefits and risks of AI. The study contributes to the growing discourse on AI governance by proposing a multi-stakeholder framework that emphasizes the importance of inclusive participation, transparency, and accountability in shaping the future of information. The research offers actionable insights for policymakers, industry leaders, and civil society organizations seeking to foster a trustworthy and inclusive information environment in the era of AI, while harnessing the potential of AI-driven innovations for sustainable development.
We apply survival analysis as well as machine learning models to predict the duration of clinical trials using the largest dataset so far constructed in this domain. Neural network-based DeepSurv yield the most accurate predictions and we identify key factors that are most predictive of trial duration. This methodology may help clinical researchers optimize trial designs for expedited testing, and can also reduce the financial risk of drug development, which in turn will lower the cost of funding and increase the amount of capital allocated to this sector.
The AI-Powered Text Synthesis Platform is an innovative web-based tool designed to enhance digital content creation through artificial intelligence. The system leverages modern web technologies such as Next.js, React, TailwindCSS, TypeScript, Gemini, and Clerk to facilitate seamless content generation. This platform serves individuals and enterprises, enabling them to generate high-quality, natural-sounding text for marketing, blogging, and business communication. Designed as a Software as a Service (SaaS) solution, it offers a scalable, efficient, and user-friendly approach to digital content production. By integrating advanced AI-driven modules, the platform significantly improves workflow efficiency and ensures competitive advantages in the evolving digital content landscape
The immersive experience of watching movies presents a unique opportunity to observe and analyze adults’ facial expressions and emotional responses in a naturalistic environment. The significant growth of the entertainment industry in recent years has expanded the availability and personalization of movies and other media, offering a platform for studying behavioral patterns in diverse populations. This study explores the feasibility of diagnosing Autism Spectrum Disorder (ASD) in adults through the analysis of facial images captured during movie watching sessions. A Kaggle dataset comprises 2653 facial images of individuals diagnosed with ASD and neurotypical adults, focusing on emotion-related expression changes over time to develop a diagnostic model. To address this challenge, we propose the integration of deep learning methodologies to classify ASD in adults aged 18 to 30 years. The data set was pre-processed to standardize and enhance the variability of the training images, including resizing to a fixed target size, horizontal flipping for augmentation and rescaling to normalize pixel values. These pre-processed images are input for a hybrid model that combines multi-convolutional neural networks (multi-CNN) and bidirectional long- and short-term memory (BiLSTM) networks. The multi-CNN component efficiently extracts spatial features from the images, while the BiLSTM leverages temporal dynamics, capturing the sequential nature of changes in facial expressions over time. The proposed model achieved an accuracy of 96.6% in predicting ASD diagnoses, demonstrating the efficacy of the work done. This high level of performance underscores the potential of integrating computer vision and deep learning techniques for non-invasive, scalable diagnostic tools. The findings contribute to the development of efficient methods for the detection of ASD in young adults, emphasizing the importance of emotion-related behavioral analysis in autism research.
Individuals with complex communication needs often rely on augmentative and alternative communication (AAC) systems to have conversations and communicate their wants. Such systems allow message authoring by arranging pictograms in sequence. However, the difficulty of finding the desired item to complete a sentence can increase as the user’s vocabulary increases. This paper proposes using BERTimbau, a Brazilian Portuguese version of Bidirectional Encoder Representations from Transformers (BERT), for pictogram prediction in AAC systems. To fine-tune BERTimbau, we constructed an AAC corpus for Brazilian Portuguese to use as a training corpus. We tested different approaches to representing a pictogram for prediction: as a word (using pictogram captions), as a concept (using a dictionary definition), and as a set of synonyms (using related terms). We also evaluated the usage of images for pictogram prediction. The results demonstrate that using embeddings computed from the pictograms’ caption, synonyms, or definitions have a similar performance. Using synonyms leads to lower perplexity, but using captions leads to the highest accuracies. This paper provides insight into how to represent a pictogram for prediction using a BERT-like model and the potential of using images for pictogram prediction.
Ambiguity is considered an indispensable attribute of all natural languages. The process of associating the precise interpretation to an ambiguous word taking into consideration the context in which it occurs is known as word sense disambiguation (WSD). Supervised approaches to WSD are showing better performance in contrast to their counterparts. These approaches, however, require sense annotated corpus to carry out the disambiguation process. This paper presents the first-ever standard WSD dataset for the Kashmiri language. The raw corpus used to develop the sense annotated dataset is collected from different resources and contains about 1 M tokens. The sense-annotated corpus is then created using this raw corpus for 124 commonly used ambiguous Kashmiri words. Kashmiri WordNet, an important lexical resource for the Kashmiri language, is used for obtaining the senses used in the annotation process. The developed sense-tagged corpus is multifarious in nature and has 19,854 sentences. Based on this annotated corpus, the Lexical Sample WSD task for Kashmiri is carried out using different machine-learning algorithms (J48, IBk, Naive Bayes, Dl4jMlpClassifier, SVM). To train these models for the WSD task, bag-of-words (BoW) and word embeddings obtained using the Word2Vec model are used. We used different standard measures, viz. accuracy, precision, recall, and F1-measure, to calculate the performance of these algorithms. Different machine learning algorithms reported different values for these measures on using different features. In the case of BoW model, SVM reported better results than other algorithms used, whereas Dl4jMlpClassifier performed better with word embeddings.
Given a series of user action sequences, Contextual Sequence-Based User Behavior Anomaly Detection (CS-UBAD) identifies anomalous sequences that deviate from normal behavior patterns. The CS-UBAD problem is important for detecting insider threats, such as unauthorized access, intellectual property theft, or other malicious activities within an organization’s systems. In this paper, we propose a novel approach called Contiguous, Contextual, and Classifying Pipeline (C3P), which integrates pattern mining and the ABC (Antecedent-Behavior-Consequence) model to calculate anomaly scores without requiring human intervention. Our method reduces the computational complexity while accurately detecting anomalous sequences.
Companies that deliver food (food delivery services, or FDS) try to use customer feedback to identify aspects where the customer experience could be improved. Consumer feedback on purchasing and receiving goods via online platforms is a crucial tool for learning about a company’s performance. Many English-language studies have been conducted on sentiment analysis (SA). Arabic is becoming one of the most extensively written languages on the World Wide Web, but because of its morphological and grammatical difficulty as well as the lack of openly accessible resources for Arabic SA, like as dictionaries and datasets, there has not been much research done on the language. Using a manually annotated FDS dataset, the current study conducts extensive sentiment analysis using reviews related to FDS that include Modern Standard Arabic and dialectal Arabic. It does this by utilizing word embedding models, deep learning techniques, and natural language processing to extract subjective opinions, determine polarity, and recognize customers’ feelings in the FDS domain. Convolutional neural network (CNN), bidirectional long short-term memory recurrent neural network (BiLSTM), and an LSTM-CNN hybrid model were among the deep learning approaches to classification that we evaluated. In addition, the article investigated different effective approaches for word embedding and stemming techniques. Using a dataset of Modern Standard Arabic and dialectal Arabic corpus gathered from Talabat.com, we trained and evaluated our suggested models. Our best accuracy was approximately 84% for multiclass classification and 92.5% for binary classification on the FDS. To verify that the proposed approach is suitable for analyzing human perceptions in diversified domains, we designed and carried out excessive experiments on other existing Arabic datasets. The highest obtained multi-classification accuracy is 88.9% on the Hotels Arabic-Reviews Dataset (HARD) dataset, and the highest obtained binary classification accuracy is 97.2% on the same dataset.
The growth of Twitter users in Indonesia is in line with the development of natural language processing (NLP) technology for text classification in various languages, including Indonesian. In Indonesia, Twitter has become one of the text-based social media platforms for voicing opinions, feelings, and complaints, including public services provided by government organizations such as the National Civil Service Agency (BKN). To support the improvement of public services, media monitoring is needed in the form of sentiment analysis that combines sentiment classification and emotions from public comments. This research aims to evaluate the performance of IndoRoBERTa model optimization in sentiment and emotion classification on Indonesian text, with public comment data on BKN Twitter as the object of study. Based on the classification results, the IndoRoBERTa-Base model showed the best performance for sentiment and emotion classification tasks. In sentiment classification, the model achieved the highest accuracy of 97.3% and F1-score of 97.0%. For emotion classification, the same model also excelled with 82.2% accuracy and 82.7% F1-score. This shows that IndoRoBERTa-Base
is very effective for Indonesian text analysis, outperforming other models. The confusion matrix evaluation shows the model’s ability to classify POS and NEG sentiments with 99% and 98% accuracy, respectively, although NEU sentiment experiences more errors with 95% accuracy. In emotion classification, the highest accuracy was found for ANG and HAP emotions with 87%, while FEA and SAD emotions showed higher error rates. The precision-recall curve analysis results show that the model achieves a micro-averaged AUC of 0.996 in sentiment classification, with an AUC of 0.996 in NEG sentiment and 0.999 in
POS. For emotion classification, the micro-averaged AUC is 0.905. On testing using BKN’s public and Twitter datasets, the model performed quite well, especially for sentiment classification with a high probability level. Overall, the results show that IndoRoBERTa-Base performs well in classifying public sentiments and emotions.
(Source:
https://repository.its.ac.id/116268/)
Natural language processing (NLP) has undergone significant changes in its methods, reflecting advances in computing technology and cognitive research. This article reviews the key stages of the evolution of natural language processing methods. The article touches on the topic of the first NLP systems developed, provides justification for the reasons for the complexity of some processed texts and the possible depth of analysis. In addition, it describes not only NLP methods before and after the GPT revolution, but also current trends and prospects in the field of natural language processing. The article allows us to trace how the idea of natural language text has changed during the development of computer analysis methods, as well as to understand what text is in the mirror of natural language processing, what is really the subject of natural language processing research and what cannot be seen through the eyes of a simple researcher who does not use NLP methods.
Computational humor recognition is considered to be one of the hardest tasks in natural language processing (NLP) since humor is such a particularly complex emotion. There are very few recent studies that offer an analysis of certain aspects of computational humor. However, there has been no attempt to study the empirical evidence on computational humor recognition in a systematic way. The aim of this research is to examine computational humor detection from three aspects: datasets, features and algorithms. Therefore, a Systematic Literature Review (SLR) was carried out to present in detail the computational techniques for humor identification under these aspects. After posing some research questions, a total of 106 primary papers were identified as relevant to the objectives of these questions and further detailed analysis was conducted. The study revealed that there are a great number of publicly available annotated humor datasets with many different types of humor instances. Twenty-one (21) humor features have been carefully studied, and research evidence of their use in humor computational detection is presented. Additionally, a classification of the humor detection approaches was performed, and the results are presented. Finally, the challenges of applying these techniques to humor recognition as well as promising future research directions are discussed.
This work investigates the evolution of traditional set theory to address complex and ambiguous real-world phenomena. It introduces hierarchical hyperstructures and superhyperstructures, where superhyperstructures are formed by iteratively applying power sets to create nested abstractions. The focus is placed on three foundational set-based frameworks-Fuzzy Sets, Neutrosophic Sets, and Plithogenic Sets-and their extensions into Hyperfuzzy Sets, HyperNeutrosophic Sets, and Hyperplithogenic Sets. These extensions are applied to various domains, including Statistics, TOPSIS, K-means Clustering, Evolutionary Theory, Topological Spaces, Decision Making, Probability, and Language Theory. By exploring these generalized forms, this paper seeks to guide and inspire further research and development in this rapidly expanding field.
The computational analysis of big data has revolutionized social science research, offering unprecedented insights into societal behaviors and trends through digital data from online sources. However, existing tools often face limitations such as technical complexity, single-source dependency, and a narrow range of analytical capabilities, hindering accessibility and effectiveness. This article introduces DataPoll, an end-to-end big data analysis platform designed to democratize computational social science research. DataPoll simplifies data collection, analysis, and visualization, making advanced analytics accessible to researchers of diverse expertise. It supports multisource data integration, innovative analytical features, and interactive dashboards for exploratory and comparative analyses. By fostering collaboration and enabling the integration of new data sources and analysis methods, DataPoll represents a significant advancement in the field. A comprehensive case study on the Ukrainian--Russian conflict demonstrates its capabilities, showcasing how DataPoll can yield actionable insights into complex social phenomena. This tool empowers researchers to harness the potential of big data for impactful and inclusive research.
Students’ preconceptions about workplace culture may influence the learning environment in the Surgery clerkship; however, the time at which students develop these sentiments is unclear. We aimed to identify inflections in students’ preconceptions of the culture of surgery, which are relevant to the timing of interventions targeting the surgical learning environment.
Cohorts of students at multiple levels received surveys between July 2021-September 2023 soliciting words associated with, “Culture of Surgery.” We analyzed entries using a Bag-of-Words method, with each word representing a unique token, and determined the most prevalent words. In sentiment analysis, 2 raters independently assigned a positive, neutral, or negative valence to each word, and valence agreement was assessed. We compared proportions of valences and rater agreement among cohorts with Fisher exact tests and determined inflections in sentiment along the learning continuum.
Participants included 50 undergraduates, 111 first-year medical students, and 216 clerks beginning Surgery rotations. “Intense” was the most common word associated with “Culture of Surgery” at all levels. Sentiment analysis comparing pre-medical undergraduate students and first-year medical students revealed profound differences in proportions of words with positive (58.9 vs 13.3%, p < 0.001) and negative (20.0 vs 57.0%, p < 0.001) valences. Non-pre-medical undergraduate students’ word valences were even more frequently positive (71.6%) and less frequently negative (7.1%). Sentiments did not change appreciably among medical students at all levels. The range of rater agreements was 62.3–78.6%.
Prior to medical school, students have predominantly positive sentiments about surgical culture, while negative preconceptions predominate early in medical school and persist into clerkships. Interventions should be designed to encourage surgeons’ contributions to all aspects of undergraduate medical education culture.
ResearchGate has not been able to resolve any references for this publication.