Reduce &amp; Attribute: Two-Step Authorship Attribution for Large-Scale Problems

Tschuggnall, Michael; Murauer, Benjamin; Specht, Günther

doi:10.18653/v1/k19-1089

Cited by 6 publications

(9 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the case of sequential sampling, the sampling window is referred to as a sliding window due to the fact that it slides across the text. It has been used in numerous authorship attribution studies [25]- [27]. The number of characters or words that the window is shifted after each iteration is known as the window step or step length [27].…”

Section: B Sample Generationmentioning

confidence: 99%

“…It has been used in numerous authorship attribution studies [25]- [27]. The number of characters or words that the window is shifted after each iteration is known as the window step or step length [27]. Therefore, a window size of 1500 words and a step length of 1500 words means that the window will pass through the text with no overlap between samples.…”

Section: B Sample Generationmentioning

confidence: 99%

See 1 more Smart Citation

Towards Performance Improvement of Authorship Attribution

Suljic,

Hossain

2024

IEEE Access

View full text Add to dashboard Cite

An accurate authorship attribution model can play a vital role in security domain by detecting fraudulent texts and combating plagiarism, online piracy, and cyber attacks. In this paper, we work on improving the performance of authorship attribution. To this end, we focus on generating effective samples and features towards creating an authorship attribution model. We did our experiments using a convolutional neural network (CNN). Two key findings from our experiments are as follows: first, our results consistently show that fusing n-grams and stylometric features yields a better performance than independently using each type of features. Notably, with fused features, we achieved an accuracy of 97.03%, a precision of 97.58%, and a recall of 97.03%. Second key finding is-when a sliding window is used in generating training samples, it is possible to improve performance by increasing the amount of overlap between samples, which can be achieved by decreasing the step length of the window. Our study shows that there is a linear relationship between performance metrics and the percent of overlap between training samples. Across three different types of features (n-grams, stylometric, and fused), the worst performance in our experiments was obtained when there was no overlap in the training samples. Inversely, the best performance was achieved when there was a 95% or a 99% overlap in the sliding windows.

show abstract

Section: B Sample Generationmentioning

confidence: 99%

Section: B Sample Generationmentioning

confidence: 99%

Towards Performance Improvement of Authorship Attribution

Suljic,

Hossain

2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…The pre-trained models like BERT, Embeddings from Language Models(ELMo), Universal Language Model Finetuning and generation Generative Pre-trained Transformer -2 www.ijacsa.thesai.org based authorship prediction on cross domain has been demonstrated in [17], a multi-headed classifier and DEMUX layer is created to handle different classifiers, BERT and ELMo outperform with more than 90% accuracy compared to other language models. Stylometry features play a vital role in the AA task, authors in [18] have explored a new technique of generating human-like sentences using a neural network and then various linguistic features are extracted to predict the authorship. the proposed model with an accuracy of 97.2% can predict the true author successfully.…”

Section: Literature Surveymentioning

confidence: 99%

Authorship Attribution on Kannada Text using Bi-Directional LSTM Technique

Chandrika¹,

Kallimani²

2022

IJACSA

View full text Add to dashboard Cite

Author attribution is the field of deducing the author of an unknown textual source based on certain characteristics inherently present in the author's style of writing. Author attribution has a ton of useful applications which help automate manual tasks. The proposed model is designed to predict the authorship of the Kannada text using a sequential neural network with Bi-Directional Long Short Term Memory layers, Dense layers, Activation function and Dropout layers. Based on the nature of the data, we have used stochastic gradient descent as an optimizer that improves the learning of the proposed model. The model extracts Part of the speech tags as one of the semantic features using the N-gram technique. A Conditional random fields model is developed to assign Part of the speech tags for the Kannada text tokens, which is the base for the proposed model. The parts of the speech model achieve an overall 90% and 91% F1 score and accuracy respectively. There is no state-of-art model to compare the performance of our model with other models developed for the Kannada language. The proposed model is evaluated using the One Versus Five (1 vs 5) method and overall accuracy of 77.8% is achieved.

show abstract

“…Supervised authorship attribution traditionally refers to the task of analyzing the linguistic patterns of a text in order to determine who, from a finite set of enrolled authors, has written a document of unknown authorship. Nowadays, the focus of this closed-set scenario has shifted from literary to social media authorship attribution, where methods have been developed to deal with large-scaled datasets of smallsized online texts (Rocha et al, 2017), (Boenninghoff et al, 2019a), (Theophilo et al, 2019), (Boenninghoff et al, 2019b) (Tschuggnall et al, 2019).…”

mentioning

confidence: 99%

Variational Autoencoder with Embedded Student-$t$ Mixture Model for Authorship Attribution

Boenninghoff¹,

Zeiler²,

Nickel³

et al. 2020

Preprint

View full text Add to dashboard Cite

Traditional computational authorship attribution describes a classification task in a closed-set scenario. Given a finite set of candidate authors and corresponding labeled texts, the objective is to determine which of the authors has written another set of anonymous or disputed texts. In this work, we propose a probabilistic autoencoding framework to deal with this supervised classification task. More precisely, we are extending a variational autoencoder (VAE) with embedded Gaussian mixture model to a Student-t mixture model. Autoencoders have had tremendous success in learning latent representations. However, existing VAEs are currently still bound by limitations imposed by the assumed Gaussianity of the underlying probability distributions in the latent space. In this work, we are extending the Gaussian model for the VAE to a Studentt model, which allows for an independent control of the "heaviness" of the respective tails of the implied probability densities. Experiments over an Amazon review dataset indicate superior performance of the proposed method.

show abstract

Reduce & Attribute: Two-Step Authorship Attribution for Large-Scale Problems

Cited by 6 publications

References 28 publications

Towards Performance Improvement of Authorship Attribution

Towards Performance Improvement of Authorship Attribution

Authorship Attribution on Kannada Text using Bi-Directional LSTM Technique

Variational Autoencoder with Embedded Student-$t$ Mixture Model for Authorship Attribution

Contact Info

Product

Resources

About