Natural Language Processing Research Articles

VOICE/TEXTUAL BOT MOBILE APP

Abstract This paper describes the design and deployment of a Voice/Textual Bot Mobile App that was aimed at revamping the way people order food using advanced speech recognition, NLP, and real-time cart management. This application allows for users to browse categorized menus and place orders in addition to being able to command and manage shopping carts through voice or text inputs. This app features secure authentication, user-friendly navigation, and real-time cart updates for users of all kinds, even the physically or technologically challenged. It bridges the gap between advanced technology and convenient everyday food delivery systems by focusing on efficiency and user satisfaction. Keywords : Voice/Textual Bot, Android Application, Speech Recognition, Natural Language Processing, Real-Time Cart Management, Food Delivery System.

Identifying Metabolic Dysfunction Associated Steatotic Liver Disease Using Natural Language Processing in a U.S National Cohort

Background and Aims: Although metabolic dysfunction-associated steatotic liver disease (MASLD) and MASLD with increased alcohol intake (MetALD) are identified as clinical entities, tools to identify patients from electronic health records (EHRs) to perform large outcome studies are lacking. Methods: In this retrospective study of participants from the Veterans Analysis of Liver Disease (VALID) cohort assembled from 1/1/2013 to 12/31/2022, a rule-based natural language processing (NLP) algorithm searched EHRs for imaging evidence of hepatic steatosis. This was combined with identification of cardiometabolic risk factors (CMRF) and harmful alcohol use. Algorithm-derived diagnoses of MASLD, MetALD, alcohol-associated steatotic liver disease (ALD), and no steatotic liver disease (SLD) were validated using a blinded review of randomly selected charts. Results: Among 817,657 eligible Veterans, SLD was present in over half (n=438,209, 53.5%), including MASLD in 299,259 (36.5%), 99,163 with MetALD (12.1%), 38,552 (4.7%) with ALD. The NLP algorithm had a high correlation with steatosis on chart review, with a kappa of 0.86 (95% CI 0.82-0.90), sensitivity of 0.96 and specificity of 0.90. Classification of MASLD, MetALD, ALD and no SLD by the algorithm also showed high correlation with chart review, with a kappa of 0.87 (95% CI 0.82-0.91). This algorithm identified 299,259 (36.5%) of the study cohort with MASLD, compared to 23,218 (2.8%) patients identified using ICD 9/10 codes. Conclusions and Relevance: An algorithm combining rule-based NLP with CMRFs and alcohol use from EHRs accurately identifies and classifies SLD and can be applied in large epidemiologic studies of SLD in the VHA.

Classifying Aviation Safety Reports: Using Supervised Natural Language Processing (NLP) in an Applied Context

This paper presents a practical approach to classifying aviation safety reports in an operational context. The goals of the research are as follows: (a) successfully demonstrate a replicable, practical methodology leveraging Natural Language Processing (NLP) to classify aviation safety report narratives; (b) determine the number of reports (per class) required to train the NLP model to achieve an F1 performance score greater than 0.90 consistently; and, (c) demonstrate the model could be implemented locally, within the confines of a typical corporate infrastructure (i.e., behind the firewall) to allay information security concerns. The authors purposefully sampled 425 safety reports from 2019 to 2021 from a university flight training program. The authors varied the number of reports used to train an NLP model to classify narrative safety reports into three separate event categories. The NLP model’s performance was evaluated both with and without distractor data, running 30 iterations at each training level. NLP model success was measured using a confusion matrix and calculating Macro Average F1-Scores. Parametric testing was conducted on macro average F1 score performance using an ANOVA and post hoc Levene statistic. We determined that 60 training samples were required to consistently achieve a macro average F1-Score above the established 0.90 performance threshold. In future studies, we intend to expand this line of research to include multi-tiered analysis to support classification within a safety taxonomy, enabling improved root cause analysis.

Artificial Intelligence-Driven High School English Reading Instruction

This study examines the integration of Artificial Intelligence (AI) technology in high school English reading instruction, analyzing its advantages over traditional teaching methods. A midst the test-oriented education system, high school English reading instruction encounters numerous challenges, including a narrow reading scope for students and a lack of diversity in teaching approaches by educators. The incorporation of AI, particularly Natural Language Processing (NLP), enables in-depth analysis of English reading comprehension and enhances teaching efficiency and student motivation through personalized recommendations, intelligent tutoring, and assessment. AI technology offers tailored learning paths based on specific student needs, achieving personalized instruction, and improves the accuracy and efficiency of assessment through intelligent evaluation and feedback mechanisms. Furthermore, AI enriches teaching resources and interaction modes, increasing student interest and engagement. The study posits that the application of AI technology in high school English reading instruction has broad prospects and is expected to significantly improve educational quality and talent cultivation.

An Embedding-Based Semantic Analysis Approach: A Preliminary Study on Redundancy Detection in Psychological Concepts Operationalized by Scales

As psychological research progresses, the issue of concept overlap becomes increasing evident, adding to participant burden and complicating data interpretation. This study introduces an Embedding-based Semantic Analysis Approach (ESAA) for detecting redundancy in psychological concepts, which are operationalized through their respective scales, using natural language processing techniques. The ESAA utilizes OpenAI’s text-embedding-3-large model to generate high-dimensional semantic vectors (i.e., embeddings) of scale items and applies hierarchical clustering to group semantically similar items, revealing potential redundancy. Three preliminary experiments evaluated the ESAA’s ability to (1) identify semantically similar items, (2) differentiate semantically distinct items, and (3) uncover overlap between scales of concepts known for redundancy issues. Additionally, comparative analyses assessed the ESAA’s robustness and incremental validity against the advanced chatbots based on GPT-4. The results demonstrated that the ESAA consistently produced stable outcomes and outperformed all evaluated chatbots. As an objective approach for analyzing relationships between concepts operationalized as scales, the ESAA holds promise for advancing research on theory refinement and scale optimization.

Fair and Ethical Resume Screening: Enhancing ATS with JustScreen the ResumeScreeningApp

Abstract— In today's fast-paced job market, the efficiency and fairness of the resume screening process are paramount. "JustScreen" emerges as a cutting-edge solution leveraging advanced Natural Language Processing (NLP) to automate resume evaluation, thus eliminating biases and promoting merit-based candidate selection. This thesis explores JustScreen's innovative approach to integrating NLP and machine learning algorithms to enhance the recruitment workflow, ensuring a more streamlined, unbiased, and efficient candidate assessment process. The methodology involves several key components: data preprocessing, NLP information extraction, fairness metrics calculation, bias mitigation, and interpretability techniques. By utilizing frameworks such as spaCy for NLP tasks, JustScreen aims to overcome the challenges of traditional manual screening processes, improving both accuracy and fairness. This thesis explores the transition from developing a full Application Tracking System (ATS) to creating a powerful enhancement for existing ATS systems. The ResumeScreeningApp/ JustScreen integrates generative AI to provide comprehensive resume analysis, adding significant value to traditional ATS functionalities. Initial evaluations indicate a significant advancement in talent acquisition practices, promoting equal opportunities and reducing the impact of potentially discriminatory factors. This research signifies a transformative shift in recruitment, setting new standards for ethical and efficient hiring practices using Generative AI.

Intellisearch - Intelligent Employee Data Access System

- The IntelliSearch Intelligent Employee Data Access System leverages advanced machine learning (ML) and natural language processing (NLP) techniques to modernize employee information retrieval. Designed to address inefficiencies in traditional manual methods, IntelliSearch provides secure, real-time responses to queries about roles, attendance, salaries, and performance metrics. Its modular architecture integrates with existing databases, offering robust data security through encryption and role-based access control (RBAC). By streamlining repetitive tasks and enhancing query accuracy, the system fosters organizational productivity and employee satisfaction. Future enhancements include predictive analytics, advanced performance tools, and industry-specific adaptations, paving the way for scalable, intelligent solutions in employee management Key Words: IntelliSearch, Machine Learning, Natural Language Processing, Employee Data Management, Role-based Access Control, Data Security, Organizational Efficiency, Real-time Queries, Scalable Architecture.

TIBW: Task-Independent Backdoor Watermarking with Fine-Tuning Resilience for Pre-Trained Language Models

Pre-trained language models such as BERT, GPT-3, and T5 have made significant advancements in natural language processing (NLP). However, their widespread adoption raises concerns about intellectual property (IP) protection, as unauthorized use can undermine innovation. Watermarking has emerged as a promising solution for model ownership verification, but its application to NLP models presents unique challenges, particularly in ensuring robustness against fine-tuning and preventing interference with downstream tasks. This paper presents a novel watermarking scheme, TIBW (Task-Independent Backdoor Watermarking), that embeds robust, task-independent backdoor watermarks into pre-trained language models. By implementing a Trigger–Target Word Pair Search Algorithm that selects trigger–target word pairs with maximal semantic dissimilarity, our approach ensures that the watermark remains effective even after extensive fine-tuning. Additionally, we introduce Parameter Relationship Embedding (PRE) to subtly modify the model’s embedding layer, reinforcing the association between trigger and target words without degrading the model performance. We also design a comprehensive watermark verification process that evaluates task behavior consistency, quantified by the Watermark Embedding Success Rate (WESR). Our experiments across five benchmark NLP tasks demonstrate that the proposed watermarking method maintains a near-baseline performance on clean inputs while achieving a high WESR, outperforming existing baselines in both robustness and stealthiness. Furthermore, the watermark persists reliably even after additional fine-tuning, highlighting its resilience against potential watermark removal attempts. This work provides a secure and reliable IP protection mechanism for NLP models, ensuring watermark integrity across diverse applications.

Research Advanced in Sarcastic Detection based on Deep Learning

Sarcasm detection is an important task in natural language processing (NLP), which aims to identify ironic intentions in text or discourse. Unlike conventional narratives, satire often uses descriptions that are opposite to the literal meaning to convey humor or criticism, and it is widely used in various scenarios such as social media, comment sections, and forums. Accurately detecting satire is crucial for understanding users' true intentions, while remains an open issue due to the freedom of language expression. Compared to face-to-face communication with the extra tone, facial expressions, and body movements, or an article with background and contextual information, a single satirical comment can hardly provide information for the machine to make a correct judgment, which may mislead the model to recognize the emotions contained in the comment. This article aims to introduce the role of different features in machine learning and deep learning satire detection tasks based on current research in this field. In addition, this article discusses the problems of satire detection and looks forward to its future development direction

Mapping the Green Urban: A Comprehensive Review of Materials and Learning Methods for Green Infrastructure Mapping

Green infrastructure (GI) plays a crucial role in sustainable urban development, but effective mapping and analysis of such features requires a detailed understanding of the materials and state-of-the-art methods. This review presents the current landscape of green infrastructure mapping, focusing on the various sensors and image data, as well as the application of machine learning and deep learning techniques for classification or segmentation tasks. After finding articles with relevant keywords, the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyzes) method was used as a general workflow, but some parts were automated (e.g., screening) by using natural language processing and large language models. In total, this review analyzed 55 papers that included keywords related to GI mapping and provided materials and learning methods (i.e., machine or deep learning) essential for effective green infrastructure mapping. A shift towards deep learning methods can be observed in the mapping of GIs as 33 articles use various deep learning methods, while 22 articles use machine learning methods. In addition, this article presents a novel methodology for automated verification methods, demonstrating their potential effectiveness and highlighting areas for improvement.

Automated analysis of spoken language differentiates multiple system atrophy from Parkinson’s disease

Background and objectivesPatients with synucleinopathies such as multiple system atrophy (MSA) and Parkinson’s disease (PD) frequently display speech and language abnormalities. We explore the diagnostic potential of automated linguistic analysis of natural spontaneous speech to differentiate MSA and PD.MethodsSpontaneous speech of 39 participants with MSA compared to 39 drug-naive PD and 39 healthy controls matched for age and sex was transcribed and linguistically annotated using automatic speech recognition and natural language processing. A quantitative analysis was performed using 6 lexical and syntactic and 2 acoustic features. Results were compared with human-controlled analysis to assess the robustness of the approach. Diagnostic accuracy was evaluated using sensitivity analysis.ResultsDespite similar disease duration, linguistic abnormalities were generally more severe in MSA than in PD, leading to high diagnostic accuracy with an area under the curve of 0.81. Compared to controls, MSA showed decreased grammatical component usage, more repetitive phrases, shorter sentences, reduced sentence development, slower articulation rate, and increased duration of pauses, whereas PD had only shorter sentences, reduced sentence development, and longer pauses. Only slower articulation rate was distinctive for MSA while unchanged for PD relative to controls. The highest correlation was found between bulbar/pseudobulbar clinical score and sentence length (r = −0.49, p = 0.002). Despite the relatively high severity of dysarthria in MSA, a strong agreement between manually and automatically computed results was achieved.DiscussionAutomated linguistic analysis may offer an objective, cost-effective, and widely applicable biomarker to differentiate synucleinopathies with similar clinical manifestations.

Bayes Theorem in Machine Learning: A Literature Review

In the era of big data and articial intelligence, machine learning has become a crucial tool for extracting insights and making predictions across various domains. Bayes theorem, a fundamental principle in probability theory, has emerged as a cornerstone in many machine learning algorithms. This literature review explores the main applications of Bayes theorem in machine learning, focusing on its role in classication, Natural Language Processing (NLP), and other emerging elds. The study aims to provide an overview of how Bayesian principles enhance learning algorithms, improve decision-making processes, and address complex problems in articial intelligence. Through a systematic review of academic papers from Google Scholar, this research synthesizes current knowledge on Bayesian methods in machine learning. The methodology involves dening the research scope, conducting a literature search using specic keywords, screening relevant studies, and analyzing the collected data. By examining diverse applications ranging from disease prediction to sentiment analysis, this review highlights the versatility and signicance of Bayes theorem in advancing machine learning techniques and their real-world implementations.

Cultural Sensitivity in AI Language Learning: Using NLP to Enhance Language Understanding Across Cultural Contexts

The aim of this paper is to demonstrate the integration of cultural awareness into AI language learning systems using Natural Language Processing (NLP) models. Recognising that the traditional methods of language learning tend not to capture cultural variations, the studies focus on how culturally relevant NLP models (GPT-3, BERT, RNNs) promote language acquisition and cross-cultural integration. Data preprocessing, cultural data augmentation and model training was used to encode cultural variables such as greetings, politeness, and contextually relevant expressions into training datasets. Experimental results suggest that culturally sensitive models outperform the generic model on tasks that involve cultural awareness, such as politeness detection, cultural phrase recognition, and tone sensitivity. GPT-3 for example improved the accuracy of politeness detection tasks from 72.4% to 85.7%. These results highlight the need to bring cultural awareness into AI to develop engaging, context-sensitive language learning. It concludes that culturally based NLP models enhance learner capacity for real-world communication and foster global cultural awareness. Future work should focus on expanding culturally diverse datasets and refining AI models to address subtle cultural complexities

Sentiment Analysis Techniques for Deep Learning Classification and Comparison

Sentiment analysis, a critical tool in natural language processing (NLP), has seen significant advancements with the rise of machine learning and deep learning techniques. This paper explores the evolution of sentiment analysis from traditional models, such as CNN and RNN, to more sophisticated approaches like LSTM, LSTM-CNN, and BERT. Using the IMDb dataset for binary classification of sentiment, we utilize metrics like F-Score, Precision, Accuracy, and Recall to compare the way these models perform. Despite the progress made, sentiment analysis still faces challenges, particularly in handling multiple languages, incorporating multimedia content, and addressing privacy and ethical concerns. This paper also discusses future directions for sentiment analysis, including the development of models for low-resource languages, real-time sentiment analysis, and the ethical implications of data handling. Overall, the study highlights the potential of sentiment analysis and the opportunities for improvement through continued research and innovation.

Machine learning in ocular oncology and oculoplasty: Transforming diagnosis and treatment

In the domains of ocular oncology and oculoplasty, machine learning (ML) has become a game-changing technology, providing previously unheard-of levels of precision in diagnosis, treatment planning, and outcome prediction. Using imaging modalities, genomic data, and clinical characteristics, this chapter investigates the integration of machine learning algorithms in the detection and treatment of ocular tumours, including retinoblastoma and uveal melanoma. Through predictive modelling and real-time decision-making, it also emphasises how ML might improve surgical outcomes in oculoplasty, including orbital reconstruction and eyelid correction. Automated examination of fundus photographs, histological slides, and 3D imaging has been made possible by methods like deep learning and natural language processing, which have improved individualised therapeutic approaches and decreased diagnostic errors. Additionally, the use of augmented reality and machine learning in robotics and surgery is a significant development in precision oculoplasty. Notwithstanding its potential, issues including data heterogeneity, algorithm interpretability, and ethical considerations are significant roadblocks that need to be addressed. This chapter explores cutting-edge developments, real-world uses, and potential future paths, offering researchers and doctors a thorough resource.Dipali Vikas Mane, Associate Professor, Shriram Shikshan Sanstha’s College of Pharmacy, Paniv-413113

A clip-based dual-stream method for text based vehicle search

A text-based vehicle search refers to a system where users can find vehicles or route information by entering text-based queries. The primary objective of text-based vehicle search is to identify the most relevant vehicle in a given dataset using a natural language description as a query. This approach leverages natural language processing (NLP) to understand and interpret description queries and provide relevant results. Despite significant progress, this task still faces several challenges due to the complexity and diversity of natural language, as well as inherent difficulties in the vision domain. Moreover, few studies have focused on tracked-vehicle retrieval, where vehicle tracklets are considered instead of single images. In this paper, we propose a novel framework for natural language-based tracked-vehicle retrieval based on CLIP model, one of the most effective models for image-text matching task. This framework leverages both appearance and motion information to enhance the matching accuracy of vehicle tracklet retrieval. Some experiments are conducted on the CityFlow-NL dataset, provided by the 6-th AI City Challenge, an annual competition. The results are comparable to state-of-the-art methods, achieving an MRR score of 46.63%, Rank@5 of 67.02%, and Rank@10 of 81.82%

Characterizing the patient experience of physical restraint in psychiatric settings via a linguistic, sentiment, and metaphor analysis

Physical Restraint (PR) is a coercive procedure used in emergency psychiatric care to ensure safety in life-threatening situations. Because of its traumatic nature, studies emphasize the importance of considering the patient’s subjective experience. We pursued this aim by overcoming classic qualitative approaches and innovatively applying a multilayered semiautomated language analysis to a corpus of narratives about PR collected from 99 individuals across seven mental health services in Italy. Compared to a reference corpus, PR narratives were characterized by reduced fluency and lexical density, yet a greater use of emotional and cognitive terms, verbs, and first-person singular pronouns. Sadness was the most represented emotion, followed by anger and fear. One-third of the PR narratives contained at least one metaphor, with Animals and War/Prison as the most distinctive source domains. The quality and length of the PR experience impacted both the structure and the sentiment of the narratives. Findings confirm the distressful nature of PR but also point to the use of various linguistic mechanisms which might serve as an early adaptive response toward healing from the traumatic experience. Overall, the study highlights the importance of Natural Language Processing as an unobtrusive window into subjective experience, offering insights for therapeutic choices.

Race and Sex Disparities in Metabolic/Bariatric Surgery over 20 Years: A Cohort Study

Objectives: To determine how rates of discussion of metabolic and bariatric surgery (MBS) between patients with class 2 obesity and higher and their healthcare providers as well as subsequent progression to MBS have varied by race and sex over the last 20 years. Background: Obesity is a growing public health crisis. MBS remains the most effective treatment for long-term management of obesity. Methods: Retrospective cohort study of electronic health records from a large tertiary academic health system using a previously validated natural language processing algorithm. The primary outcomes were discussion of MBS between eligible patients and providers and progression to surgery. Multivariable logistic regression was used to determine the association between sociodemographic factors, time, and discussion and receipt of MBS. Results: Among 122,487 patients included in the analysis, 11,094 (9.1%) patients discussed MBS with their providers. Of these, 1348 (12.2%) progressed to MBS. Between 2000 and 2020 the annual fraction of patients who had a discussion of MBS with their providers rose from 3.2% to 10.0% (P < 0.001). Black people were equally likely to discuss MBS with providers as non-Black people (9.5 vs 9.0%, P = 0.09) but were less likely to progress to MBS (8.4 vs 12.6%, P < 0.001). These disparities in progression narrowed over the study period (P = 0.044). Men were less likely than women to both discuss (8.1% vs 9.6%, P < 0.001) and receive MBS (7.6% vs 14.6%, P < 0.001), and these disparities increased over the study period. Conclusions: Interventions to reduce racial disparities in utilization of MBS should focus on progression from discussion to MBS. Interventions to increase the uptake of MBS among men are required across the care cascade.

The Practice of Large Language Models in Automated Question Generation: A Case Study of ChatGLM in High School Information Technology Curriculum

With the advancement of information technology, the field of education is undergoing transformation. In the teaching of high school information technology, scientific and efficient question formulation is crucial for enhancing the quality of education. Traditional methods of question formulation rely on the experience of teachers, leading to issues such as inconsistent difficulty levels and incomplete coverage of knowledge points. Large language models (LLMs), such as ChatGLM, leverage natural language processing and deep learning technologies to automatically generate questions that align with the curriculum, thereby improving scientific accuracy and precision, enhancing diversity and innovation, and catering to students’ personalized needs. Additionally, LLMs can enhance the efficiency of question formulation and reduce the burden on teachers. This paper explores the application value of large language models in the formulation of questions for high school information technology, through empirical research comparing the performance of human-generated and ChatGLM-generated questions in terms of accuracy, relevance, clarity, and willingness. The study selected two chapters, “Data and Information” and “Fundamentals of Algorithms,” and employed both human and ChatGLM-generated questions, inviting teachers to evaluate them. Through data analysis and statistical testing, we reveal the advantages and limitations of large language models in educational question formulation, providing insights for the intelligent development of educational assessment systems.

Artificial Intelligence in Library Studies

Artificial intelligence has emerged as a promising technology in the post-pandemic era, significantly impacting the library ecosystem and the direction of library and information science studies. This study aims to map AI-related research in libraries to identify opportunities and discuss future directions. Using textual analysis of data from Scopus, the study analyzed article titles with the burst detection algorithm and abstracts with scattertext and lemmatization. Six burst words were detected out of twelve frequently appearing in titles. Scattertext results showed a comparison between the service side (red) and the development side (blue) in libraries. Research increasingly focuses on AI utilization for library services and natural language processing (NLP) to enhance services. On the development side, AI involves product creation and encompasses AI literacy frameworks, policies, and their impact on libraries. AI affects studies in libraries by changing application methods, such as machine learning and NLP. Future research will become more diverse, considering the unique characteristics of each library.