Password region attribute classification based on multi-granularity cascade fusion.
Saved in:
| Title: | Password region attribute classification based on multi-granularity cascade fusion. |
|---|---|
| Authors: | Yu, Wei (AUTHOR), Liu, Cheng (AUTHOR), Ni, Lvlin (AUTHOR), Shi, Yu (AUTHOR), Ji, Qingbing (AUTHOR) |
| Source: | Connection Science. Dec 2025, Vol. 37 Issue 1, p1-25. 25p. |
| Subjects: | Classification, Machine learning, Polysemy |
| Abstract: | The composition of the password is markedly disparate contingent on the configuration strategy and the individual user's predilections. The objective of this paper is to mine the region attribute information behind the password text through text classification. In contrast to the traditional text classification approach, the classification of password region attribution represents a distinct challenge namely ultra-short text classification. The issue of password regional attribute classification is particularly tricky due to its inherent lexical polysemy, the scarcity of text features, the lack of context and the difficulty in explicitly identifying semantics. To address the aforementioned issues, we propose a multi-granularity cascade fusion approach for password region attribution classification. Firstly, the model employs series of segmentation techniques to split password into multi-dimensional fine-grained subword representations. Subsequently, multiple segmented representations of the same password are fed into a localised feature encoder to mine the private local features. Finally, a multi-level cascade fusion method is designed to integrate different granularity of password features into a unified representation to classification. Our approach can effectively addresses the limitations of scarce information and the challenge of integrating multiple representations for password text. Experiments on a large amount of real password data demonstrate that, our model can converge rapidly and achieve an accuracy of 88.18%, a precision of 88.31%, a recall of 87.73%, and an F1-score of 88.02%, significantly outperforming traditional models. [ABSTRACT FROM AUTHOR] |
| Copyright of Connection Science is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Psychology and Behavioral Sciences Collection |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | The composition of the password is markedly disparate contingent on the configuration strategy and the individual user's predilections. The objective of this paper is to mine the region attribute information behind the password text through text classification. In contrast to the traditional text classification approach, the classification of password region attribution represents a distinct challenge namely ultra-short text classification. The issue of password regional attribute classification is particularly tricky due to its inherent lexical polysemy, the scarcity of text features, the lack of context and the difficulty in explicitly identifying semantics. To address the aforementioned issues, we propose a multi-granularity cascade fusion approach for password region attribution classification. Firstly, the model employs series of segmentation techniques to split password into multi-dimensional fine-grained subword representations. Subsequently, multiple segmented representations of the same password are fed into a localised feature encoder to mine the private local features. Finally, a multi-level cascade fusion method is designed to integrate different granularity of password features into a unified representation to classification. Our approach can effectively addresses the limitations of scarce information and the challenge of integrating multiple representations for password text. Experiments on a large amount of real password data demonstrate that, our model can converge rapidly and achieve an accuracy of 88.18%, a precision of 88.31%, a recall of 87.73%, and an F1-score of 88.02%, significantly outperforming traditional models. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 09540091 |
| DOI: | 10.1080/09540091.2025.2461092 |
