Article
Biochemistry & Molecular Biology
John A. Bachman, Benjamin M. Gyori, Peter K. Sorger
Summary: This study presents an approach to accurately assemble molecular mechanisms by using multiple natural language processing systems and INDRA, which improves the reliability of machine reading and assembles non-redundant mechanistic knowledge. Through this approach, the study extends protein-protein interaction databases and provides explanations for co-dependencies in the Cancer Dependency Map.
MOLECULAR SYSTEMS BIOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
Martin Perez-Perez, Tania Ferreira, Gilberto Igrejas, Florentino Fdez-Riverola
Summary: Discovering relevant biomedical interactions is crucial for biology research. This study proposes a novel vector-space integrated with a deep learning model to assist manual curators in a real curation task. Experimental results show that the proposed workflow is valuable for semi-automatic relation extraction and saves manual annotation efforts.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Engineering, Environmental
Emma H. Palm, Parviel Chirsir, Jessy Krier, Paul A. Thiessen, Jian Zhang, Evan E. Bolton, Emma L. Schymanski
Summary: Transformation product (TP) information is crucial for assessing the hazards of compounds, but the availability and usability of TP data are often limited. FAIRifying existing TP knowledge can improve data accessibility for identification workflows. ShinyTPs is an application that curates and visualizes text-mined chemical names to validate automatically extracted reactions. The application was successful in retrieving and adding newly curated reactions to the PubChem Transformations library, supporting TP identification in non-target analysis workflows.
ENVIRONMENTAL SCIENCE & TECHNOLOGY LETTERS
(2023)
Article
Computer Science, Artificial Intelligence
Mantas Vaskevicius, Jurgita Kapociute-Dzikiene, Arnas Vaskevicius, Liudas Slepikas
Summary: This article proposes a methodology that uses machine learning algorithms to extract actions from structured chemical synthesis procedures. The proposed pipeline combines ML algorithms and scripts to extract relevant data from patents, helping transform experimental procedures into structured actions. The developed pipeline enables the creation of a dataset of chemical reactions and their procedures in a structured format, facilitating the application of AI-based approaches to streamline synthetic pathways and optimize experimental conditions.
PEERJ COMPUTER SCIENCE
(2023)
Article
Computer Science, Information Systems
Xie Renchunzi, Mahardhika Pratama
Summary: Despite the challenging problem of knowledge transfer across multiple streaming processes, the proposed automatic online multi-source domain adaptation (AOMSDA) technique effectively addresses the issue by integrating a central moment discrepancy (CMD)-based regularizer under a coupled generative and discriminative approach of denoising autoencoder (DAE). Numerical studies show that AOMSDA outperforms its counterparts in 5 out of 8 cases, with ablation studies highlighting the advantages of each learning component. AOMSDA is also generalizable for any number of source streams and the source code is publicly available.
INFORMATION SCIENCES
(2022)
Article
Biotechnology & Applied Microbiology
Erica L. Lyons, Daniel Watson, Mohammad S. Alodadi, Sharie J. Haugabook, Gregory J. Tawa, Fady Hannah-Shmouni, Forbes D. Porter, Jack R. Collins, Elizabeth A. Ottinger, Uma S. Mudunuri
Summary: Rare diseases are difficult to diagnose and treat. Genetic sequencing has the potential to improve the diagnostic process, but there are challenges in interpreting variant pathogenicity and communicating known causative variants. This study investigated the translation of variant knowledge from published manuscripts to public databases and found that some pathogenic variants were inaccessible, limiting the use of this information for diagnosis and treatment. Developing text mining workflows that combine natural language processing and impact prediction algorithms could be a promising approach to address this issue.
Review
Computer Science, Artificial Intelligence
Megha Chaudhary, Divya Bansal
Summary: In the contemporary era, terrorists have taken advantage of the extensive use of the internet and social media to spread their propaganda and further their goals. Open Source Intelligence (OSINT) provides a solution to analyze online information and extract intelligence in combating terrorism. This paper reviews the latest developments in OSINT and discusses tools and techniques for extracting terrorism-related textual information from publicly accessible sources. It also highlights the challenges and gaps in different phases of OSINT extraction.
WILEY INTERDISCIPLINARY REVIEWS-DATA MINING AND KNOWLEDGE DISCOVERY
(2022)
Article
Computer Science, Artificial Intelligence
Sahar Behpour, Mohammadmahdi Mohammadi, Mark V. Albert, Zinat S. Alam, Lingling Wang, Ting Xiao
Summary: The study demonstrates the importance of emphasizing time in trend detection by introducing a weighted temporal feature. By analyzing finance journal abstracts, trending finance topics that are not identifiable with standard clustering methods are discovered. The use of silhouette score divided by standard deviation to identify and validate trending topics showcases the effectiveness of the approach.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
H. Abo-Bakr, S. A. Mohamed
Summary: Due to the vast amount of textual information, automatic text summarization (ATS) systems are necessary for extracting important information or generating summaries. This work proposes an extractive ATS system that preprocesses the text and formulates the summarization as a multi-objective optimization problem. An evolutionary sparse multi-objective algorithm is developed to solve the problem and produce a set of non-dominated summaries. The system has been evaluated using DUC datasets and compared to existing literature using ROUGE metrics.
COMPLEX & INTELLIGENT SYSTEMS
(2023)
Review
Computer Science, Artificial Intelligence
Gyunam Park, Minsu Cho, Jiyoon Lee
Summary: This article provides an in-depth analysis of process mining using text mining and machine learning techniques, including main research fields, relationships between fields, and future development trends.
EXPERT SYSTEMS WITH APPLICATIONS
(2024)
Article
Biochemistry & Molecular Biology
Paola Turina, Piero Fariselli, Emidio Capriotti
Summary: In recent years, the increase in DNA sequencing and protein mutagenesis studies has generated a large amount of variation data. The manual curation of data from literature is time-consuming and costly, prompting the development of tools like ThermoScan for extracting relevant thermodynamic data on protein stability from full-text articles. ThermoScan's text mining approach has shown accurate predictions and outperformed other text-mining algorithms based on publication abstracts.
FRONTIERS IN MOLECULAR BIOSCIENCES
(2021)
Article
Green & Sustainable Science & Technology
Zuqi Wang, Yanting Qin
Summary: The study of public attitude towards the Shanghai epidemic is of great importance for maintaining mental health and a positive outlook. This study utilized crawler technology to obtain Weibo data related to the epidemic in Shanghai and classified attitudes using sentiment dictionaries. The results showed that the overall attitude of the public in Shanghai is positive, although there were fluctuations in the early stages of the epidemic. Public concerns primarily focused on the supply of goods and virus testing. Public attitudes in areas close to the epicenter were relatively more negative.
Article
Urban Studies
Valentina Marchi, Alessandra Marasco, Valentina Apicerni
Summary: This study introduces a text-mining approach to analyze the communication of sustainability on websites of 10 European tourism cities. The results show that this approach can automatically assess the depth and balance of communication across different dimensions of sustainability.
Article
Psychology, Multidisciplinary
Wei Wang, Ling He, Yenchun Jim Wu, Mark Goh
Summary: This study examines the influence of subjectivity versus objectivity in crowdfunding narratives on fundraising outcomes. By employing text mining and Bayesian inference, the study identifies and differentiates subjective and objective attributes in entrepreneurial narratives. The research suggests that a strategic positioning of subjective and objective claims in different text sections can enhance the success of online fundraising, resulting in an improved prediction accuracy of funding support.
COMPUTERS IN HUMAN BEHAVIOR
(2021)
Article
Cardiac & Cardiovascular Systems
Benjamin Neumann, A. Suzanne Vink, Ben J. M. Hermans, Krystien V. V. Lieve, Didem Comert, Britt-Maria Beckmann, Sally-Ann B. Clur, Nico A. Blom, Tammo Delhaas, Arthur A. M. Wilde, Stefan Kaeaeb, Pieter G. Postema, Moritz F. Sinner
Summary: Pairwise comparisons between automatic and manual measurements of QT intervals and QTc revealed significant discrepancies, especially in patients with aberrant QTc values. Novel automatic algorithms may be able to improve the agreement between these measurements.