Article
Computer Science, Artificial Intelligence
Haiyan Wu, Zhiqiang Zhang, Qingfeng Wu
Summary: This paper discusses the importance of authorship attribution and the limitations of existing methods, proposing a novel approach that combines features from multiple dimensions, with experimental results demonstrating its effectiveness compared to state-of-the-art models.
APPLIED SOFT COMPUTING
(2021)
Article
Computer Science, Information Systems
Andreas Neocleous, Antis Loizides
Summary: This article revisits a divisive issue regarding the authorship of John Stuart Mill's corpus, analyzing experts' differing opinions and the research team's methods and experimental results. By training classifiers, disputed texts are attributed to John Stuart Mill.
Article
Multidisciplinary Sciences
Hend Alrasheed
Summary: Keyword extraction involves detecting the most relevant terms and expressions in text. Using graph analysis tools for keyword extraction to assess topic diversity and sentiment within the text.
Article
Computer Science, Information Systems
Ai Zhou, Yijia Zhang, Mingyu Lu
Summary: This study proposes an approach to analyze the authorship of classical Chinese poetry, by evaluating the popularity of poets and building a public corpus for authorship profiling. A novel framework named M-DKPP is proposed, which combines authorship attribution knowledge, text's stylistic features, and domain knowledge from experts in traditional poetry studies. The validity and applicability of the framework are demonstrated through a case study on Li Bai, and its performance is evaluated on four poem datasets, outperforming several baseline approaches for authorship attribution.
Article
Computer Science, Theory & Methods
Jose Antonio Garcia-Diaz, Ricardo Colomo-Palacios, Rafael Valencia-Garcia
Summary: This study investigates the reliability of determining psychographic traits concerning political ideology and presents the PoliCorpus-2020 dataset for authorship analysis tasks. The results show that linguistic features are effective indicators for identifying political affiliation and improve the performance of neural network models.
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE
(2022)
Article
Computer Science, Artificial Intelligence
Rui Shao, Pramuditha Perera, Pong C. Yuen, Vishal M. Patel
Summary: This paper studies two critical aspects of deep learning: open-set recognition and adversarial defense. It finds that open-set recognition systems are vulnerable to adversarial samples, and adversarial defense mechanisms trained on known classes are ineffective for open-set samples. Based on these findings, the paper proposes an Open-Set Defense Network with Clean-Adversarial Mutual Learning (OSDN-CAML) to address this problem.
INTERNATIONAL JOURNAL OF COMPUTER VISION
(2022)
Review
Multidisciplinary Sciences
Leonard Bauersfeld, Angel Romero, Manasi Muglikar, Davide Scaramuzza
Summary: In this study, a transformer-based neural network architecture is proposed to attribute an anonymous manuscript to an author using only the text content and author names. The largest authorship-identification dataset to date was created by leveraging over 2 million publicly available research papers on arXiv. The method achieves an unprecedented authorship attribution accuracy, correctly attributing up to 73% of papers in subsets with up to 2,000 different authors.
Article
Computer Science, Artificial Intelligence
Caio Deutsch, Ivandre Paraboni
Summary: Authorship attribution and author profiling are two related fields that can benefit from each other. This paper improves the authorship attribution model by adding author demographics predictions, and evaluates the enriched model in different domains and languages, showing better performance compared to the standard method.
NATURAL LANGUAGE ENGINEERING
(2023)
Article
Mathematics
Gabriela Czibula, Mihaiela Lupea, Anamaria Briciu
Summary: This paper discusses the problem of code authorship attribution and introduces the AutoSoft model for identifying developers based on their programming style. The model, built using autoencoders, shows superior performance in various test settings compared to existing solutions. AutoSoft not only outperforms other methods in code authorship attribution, but also offers adaptability and extensions.
Article
Medicine, Legal
Patrick Juola
Summary: The paper introduces a computer program to identify the author of anonymous or disputed documents, and validates its accuracy through a series of controlled experiments involving English language blogs. The system achieved a measured accuracy of 77% across over 32,000 different document pairs, providing a solution to a key problem in forensic linguistics.
FORENSIC SCIENCE INTERNATIONAL
(2021)
Article
Computer Science, Information Systems
Andreas Neocleous, Giorgos Kataliakos, Antis Loizides
Summary: This study uses machine learning techniques to investigate the authorship of two famous essays in the nineteenth century. The classifiers trained in this research show that John Stuart Mill is the primary author of the essays, but also highlight the contribution of Harriet Taylor Mill to certain portions of text.
Article
Computer Science, Artificial Intelligence
Fatimah Alqahtani, Mischa Dohler
Summary: Authorship identification is the process of analyzing writing styles to determine the author's identity, which is important in digital forensics and cyber investigations. This survey focuses on authorship identification in the Arabic language and reviews 27 prominent studies, considering data, features, methods, and results. The results vary based on features and datasets, and challenges are faced in data preprocessing due to the complexities of Arabic morphology.
ACM TRANSACTIONS ON ASIAN AND LOW-RESOURCE LANGUAGE INFORMATION PROCESSING
(2023)
Article
Computer Science, Theory & Methods
Roni Mateless, Oren Tsur, Robert Moskovitch
Summary: This paper introduces a novel approach for software package authorship attribution called Pkg2Vec, based on a hierarchical deep neural network architecture, which better reflects real-world scenarios where code is organized in packages and written by teams. By utilizing a hierarchical neural network model and resilient features like keywords and API calls, Pkg2Vec outperforms other approaches in a large dataset of public packages.
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE
(2021)
Article
Computer Science, Artificial Intelligence
Jianting Chen, Shuhan Yuan, Dongdong Lv, Yang Xiang
Summary: Feature selection plays a crucial role in improving the accuracy and generalization of machine learning models, especially for high-dimensional data tasks. In this study, a novel self-learning feature selection approach based on feature attributions was proposed, showing improved search efficiency for optimal feature subset selection. Experimental results demonstrated the effectiveness of the SLFS approach in achieving optimal subsets with fewer iterations and utilizing SHAP values for enhanced search efficiency.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Medicine, General & Internal
Lisa M. DeTora
Summary: Applying authorship criteria in complex situations can be challenging. Existing guidelines emphasize intellectual input and accountability, while contributor taxonomies list additional activities that should be credited. However, no publication has mapped specific authorship criteria to contributor taxonomies. Suggestions are needed to differentiate activities that meet author criteria from other contributions outlined in existing taxonomies.
CURRENT MEDICAL RESEARCH AND OPINION
(2022)