Using large language models for extracting and pre-annotating texts on mental health from noisy data in a low-resource language DOI Creative Commons
Sergei Koltcov,

Anton Surkov,

Olessia Koltsova

et al.

PeerJ Computer Science, Journal Year: 2024, Volume and Issue: 10, P. e2395 - e2395

Published: Nov. 28, 2024

Recent advancements in large language models (LLMs) have opened new possibilities for developing conversational agents (CAs) various subfields of mental healthcare. However, this progress is hindered by limited access to high-quality training data, often due privacy concerns and high annotation costs low-resource languages. A potential solution create human-AI systems that utilize extensive public domain user-to-user user-to-professional discussions on social media. These discussions, however, are extremely noisy, necessitating the adaptation LLMs fully automatic cleaning pre-classification reduce human effort. To date, research LLM-based health scarce. In article, we explore zero-shot classification using four select pre-classify texts into topics representing psychiatric disorders, order facilitate future development CAs disorder-specific counseling. We use 64,404 Russian-language from online discussion threads labeled with seven most commonly discussed disorders: depression, neurosis, paranoia, anxiety disorder, bipolar obsessive-compulsive borderline personality disorder. Our shows while preliminary data filtering technology slightly improves classification, LLM fine-tuning makes a far larger contribution its quality. Both standard natural inference (NLI) modes increase accuracy more than three times compared non-fine-tuned preliminarily filtered data. Although NLI achieves higher (0.64) approach, it six slower, indicating need further experimentation hypothesis engineering. Additionally, demonstrate lemmatization does not affect quality multilingual their original perform better English-only automatically translated texts. Finally, introduce our dataset model as first openly available resource

Language: Английский

Contextual Hypergraph Networks for Enhanced Extractive Summarization: Introducing Multi-Element Contextual Hypergraph Extractive Summarizer (MCHES) DOI Creative Commons
Aytuğ Onan, Hesham Alhumyani

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(11), P. 4671 - 4671

Published: May 29, 2024

Extractive summarization, a pivotal task in natural language processing, aims to distill essential content from lengthy documents efficiently. Traditional methods often struggle with capturing the nuanced interdependencies between different document elements, which is crucial producing coherent and contextually rich summaries. This paper introduces Multi-Element Contextual Hypergraph Summarizer (MCHES), novel framework designed address these challenges through an advanced hypergraph-based approach. MCHES constructs contextual hypergraph where sentences form nodes interconnected by multiple types of hyperedges, including semantic, narrative, discourse hyperedges. structure captures complex relationships maintains narrative flow, enhancing semantic coherence across summary. The incorporates Homogenization Module (CHM), harmonizes features diverse Attention (HCA), employs dual-level attention mechanism focus on most salient information. innovative Read-out Strategy selects optimal set compose final summary, ensuring that latter reflects core themes logical original text. Our extensive evaluations demonstrate significant improvements over existing methods. Specifically, achieves average ROUGE-1 score 44.756, ROUGE-2 24.963, ROUGE-L 42.477 CNN/DailyMail dataset, surpassing best-performing baseline 3.662%, 3.395%, 2.166% respectively. Furthermore, BERTScore values 59.995 CNN/DailyMail, 88.424 XSum, 89.285 PubMed, indicating superior alignment human-generated Additionally, MoverScore 87.432 60.549 59.739 highlighting its effectiveness maintaining movement ordering. These results confirm sets new standard for extractive summarization leveraging hypergraphs better thematic fidelity.

Language: Английский

Citations

7
Machine learning‐enabled risk prediction of self‐neglect among community‐dwelling older adults in China DOI
Tengfei Li,

Yuan Xu,

Jianwei Li

et al.

Psychogeriatrics, Journal Year: 2025, Volume and Issue: 25(1)

Published: Jan. 1, 2025

Abstract Background Elder self‐neglect (ESN) is usually ignored as a private problem and impairs the health outcomes of older adults. It essential to construct robust efficient tool for risk prediction which can better detect prevent among Methods This study included 2494 participants from Ma'anshan Healthy Ageing Cohort (MHAC). First, group‐based trajectory model (GBTM) was used estimate ESN development groups. Then, feature selection methods were select variables; after that, we compared six machine learning models (Decision Tree Classifier (DT), K‐Nearest Neighbour (KNN), Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM) XGBoost (XGB)). In addition, Synthetic Minority Oversampling Technique (SMOTE) address data imbalance problem. Results The results show that be defined two groups (rising stable). After selection, final contains eight predictors. area under curve (AUC) raw dataset 0.637–0.769. with SMOTE, AUC 0.635–0.765 RF optimal model. top five most important characteristics quality life, psychological resilience, social support, education, income. Conclusions developed in this may considered simple scientific aid community‐dwelling old

Language: Английский

Citations

0
Clever Hans in the Loop? A Critical Examination of ChatGPT in a Human-in-the-Loop Framework for Machinery Functional Safety Risk Analysis DOI Creative Commons
Padma Iyenghar

Eng—Advances in Engineering, Journal Year: 2025, Volume and Issue: 6(2), P. 31 - 31

Published: Feb. 7, 2025

This paper presents a first-of-its-kind evaluation of integrating Large Language Models (LLMs) within Human-In-The-Loop (HITL) framework for risk analysis in machinery functional safety, adhering to ISO 12100. The methodology systematically addresses LLM limitations, such as hallucinations and lack domain-specific expertise, by embedding expert oversight ensure reliable compliant outputs. Applied four diverse industrial case studies—motorized gates, autonomous transport vehicles, weaving machines, rotary printing presses—this study assesses the applicability ChatGPT routine tasks central safety workflows, hazard identification assessment. results demonstrated substantial improvements: during HITL involvement subsequent iterations assessment with feedback, complete agreement ground truth was achieved across all use cases. also identified additional scenarios edge cases, enriching analysis. Efficiency gains were notable, time efficiency rated at 4.95 out 5, on average, studies. Overall accuracy (4.7 5) usability (4.8 ratings robustness ensuring practical Likert scale evaluations reflected high confidence refined outputs, emphasizing critical role enhancing both trust usability. highlights importance prompt design, revealing that longer initial prompts improve accuracy, while shorter iterative maintain without compromising efficiency. process further ensures outputs align standards requirements. underscores transformative potential generative AI activities rigorous human safety-critical, regulated industries.

Language: Английский

Citations

0
Gender Biases within Artificial Intelligence and ChatGPT: Evidence, Sources of Biases and Solutions DOI Creative Commons

Jerlyn Q.H. Ho,

Andree Hartanto,

Andrew Koh

et al.

Computers in Human Behavior Artificial Humans, Journal Year: 2025, Volume and Issue: unknown, P. 100145 - 100145

Published: March 1, 2025

Language: Английский

Citations

0
Nano-ESG: Extracting Corporate Sustainability Information from News Articles DOI

Fabian Billert,

Stefan Conrad

Lecture notes in computer science, Journal Year: 2025, Volume and Issue: unknown, P. 324 - 338

Published: Jan. 1, 2025

Language: Английский

Citations

0
An improved naive Bayes algorithm based on k ′ k -means reclassification algorithm for imbalanced classification DOI
Yanfeng Zhang, Li‐Chun Wang, Xin Wang

et al.

Communications in Statistics - Simulation and Computation, Journal Year: 2025, Volume and Issue: unknown, P. 1 - 13

Published: April 26, 2025

Language: Английский

Citations

0
Arabic Fake News Detection Using Deep Learning DOI Creative Commons
Nermin Abdelhakim Othman, Doaa S. Elzanfaly, Mostafa Mahmoud M. Elhawary

et al.

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 122363 - 122376

Published: Jan. 1, 2024

Language: Английский

Citations

3
Human-computer interactions with farm animals—enhancing welfare through precision livestock farming and artificial intelligence DOI Creative Commons
Suresh Neethirajan, Stacey D. Scott, Clara Mancini

et al.

Frontiers in Veterinary Science, Journal Year: 2024, Volume and Issue: 11

Published: Nov. 14, 2024

While user-centered design approaches stemming from the human-computer interaction (HCI) field have notably improved welfare of companion, service, and zoo animals, their application in farm animal settings remains limited. This shortfall has catalyzed emergence animal-computer (ACI), a discipline extending technology’s reach to multispecies user base involving both animals humans. Despite significant strides other sectors, adaptation HCI ACI (collectively HACI) welfare—particularly for dairy cows, swine, poultry—lags behind. Our paper explores potential HACI within precision livestock farming (PLF) artificial intelligence (AI) enhance individual address unique challenges these settings. It underscores necessity transitioning productivity-focused animal-centered methods, advocating paradigm shift that emphasizes as integral sustainable practices. Emphasizing ‘One Welfare’ approach, this discussion highlights how integrating technologies not only benefits health, productivity, overall well-being but also aligns with broader societal, environmental, economic benefits, considering pressures farmers face. perspective is based on insights one-day workshop held June 24, 2024, which focused advancing welfare.

Language: Английский

Citations

3
Enhancing risk and crisis communication with computational methods: A systematic literature review DOI Creative Commons

Madison H. Munro,

Ross Gore, Christopher J. Lynch

et al.

Risk Analysis, Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 15, 2024

Abstract Recent developments in risk and crisis communication (RCC) research combine social science theory data tools to construct effective messages efficiently. However, current systematic literature reviews (SLRs) on RCC primarily focus computationally assessing message efficacy as opposed efficiency. We conduct an SLR highlight any computational methods that improve construction found most focuses using theoretical frameworks analyze or classify elements efficacy. For improving efficiency, manual are only used classification. Specifying the is sparse. recommend future apply toward efficiency construction. By messaging would quickly warn better inform affected communities impacted by hazards. Such has potential save many lives possible.

Language: Английский

Citations

1
Toward Crops Prediction in Indonesia DOI
Prima Wahyu Titisari, Arbi Haza Nasution,

Elfis Elfis

et al.

Lecture notes in networks and systems, Journal Year: 2024, Volume and Issue: unknown, P. 207 - 216

Published: Jan. 1, 2024

Language: Английский

Citations

0