Domain Specific Author Attribution based on Feedforward Neural Network Language Models

Ge, Zhenhao; Sun, Yufang

doi:10.5220/0005710005970604

Cited by 12 publications

(13 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Different ParChoice-variants performed comparably. 19 Unlike A 4 NT, ParChoice achieved high scores (∼ 50) on both large and small datasets. We also compared METEOR scores with each Par-Choice-module applied alone in the ParChoice-LSTM variant.…”

Section: Semantic Retainmentmentioning

confidence: 98%

“…Author attribution via stylometry has traditionally focused on standard machine learning (ML) algorithms and feature engineering [1,22,33,56,68,74], but deep learning methods have become more prominent in recent years [4,9,19,67]. While there is no unanimous agreement on the most effective features [22,24,33], the Writeprints feature set has been widely applied with success [1-3, 15, 47, 53, 74].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Effective writing style transfer via combinatorial paraphrasing

Gröndahl

Asokan

2020

Proceedings on Privacy Enhancing Technologies

View full text Add to dashboard Cite

Stylometry can be used to profile or deanonymize authors against their will based on writing style. Style transfer provides a defence. Current techniques typically use either encoder-decoder architectures or rule-based algorithms. Crucially, style transfer must reliably retain original semantic content to be actually deployable. We conduct a multifaceted evaluation of three state-of-the-art encoder-decoder style transfer techniques, and show that all fail at semantic retainment. In particular, they do not produce appropriate paraphrases, but only retain original content in the trivial case of exactly reproducing the text. To mitigate this problem we propose ParChoice: a technique based on the combinatorial application of multiple paraphrasing algorithms. ParChoice strongly outperforms the encoder-decoder baselines in semantic retainment. Additionally, compared to baselines that achieve nonnegligible semantic retainment, ParChoice has superior style transfer performance. We also apply ParChoice to multi-author style imitation (not considered by prior work), where we achieve up to 75% imitation success among five authors. Furthermore, when compared to two state-of-the-art rule-based style transfer techniques, ParChoice has markedly better semantic retainment. Combining ParChoice with the best performing rulebased baseline (Mutant-X [34]) also reaches the highest style transfer success on the Brennan-Greenstadt and Extended-Brennan-Greenstadt corpora, with much less impact on original meaning than when using the rulebased baseline techniques alone. Finally, we highlight a critical problem that afflicts all current style transfer techniques: the adversary can use the same technique for thwarting style transfer via adversarial training. We show that adding randomness to style transfer helps to mitigate the effectiveness of adversarial training.

show abstract

Section: Semantic Retainmentmentioning

confidence: 98%

Section: Introductionmentioning

confidence: 99%

Effective writing style transfer via combinatorial paraphrasing

Gröndahl

Asokan

2020

Proceedings on Privacy Enhancing Technologies

View full text Add to dashboard Cite

show abstract

“…They reported an accuracy level of 76.1% using a corpus written by 50 authors where each author has 1,000 samples. Moreover, Ge [8] performed authorship identifcation using the feedforward neural networks. The authors from [8] reported 95% accuracy and noted that this task is too easy to perform due to the availability of huge training data.…”

Section: Deep Learning Based Methods To Authorship Identificationmentioning

confidence: 99%

“…The word tokenization is performed using DeepCut 7 . The rest of the features are calculated using Thai Language Toolkit (TLTK) 8 . We provide an example of Thai text sample in Table 2 and show the computed stylometric feature values in Table 3.…”

Section: Preprocessingmentioning

confidence: 99%

StyloThai:

Sarwar

Porthaveepong

Rutherford

et al. 2020

ACM Trans. Asian Low-Resour. Lang. Inf. Process.

View full text Add to dashboard Cite

Authorship identification helps to identify the true author of a given anonymous document from a set of candidate authors. The applications of this task can be found in several domains, such as law enforcement agencies and information retrieval. These application domains are not limited to a specific language, community, or ethnicity. However, most of the existing solutions are designed for English, and a little attention has been paid to Thai. These existing solutions are not directly applicable to Thai due to the linguistic differences between these two languages. Moreover, the existing solution designed for Thai is unable to (i) handle outliers in the dataset, (ii) scale when the size of the candidate authors set increases, and (iii) perform well when the number of writing samples for each candidate author is low. We identify a stylometric feature space for the Thai authorship identification task. Based on our feature space, we present an authorship identification solution that uses the probabilistic k nearest neighbors classifier by transforming each document into a collection of point sets. Specifically, this document transformation allows us to (i) use set distance measures associated with an outlier handling mechanism, (ii) capture stylistic variations within a document, and (iii) produce multiple predictions for a query document. We create a new Thai authorship identification corpus containing 547 documents from 200 authors, which is significantly larger than the corpus used by the existing study (an increase of 32 folds in terms of the number of candidate authors). The experimental results show that our solution can overcome the limitations of the existing solution and outperforms all competitors with an accuracy level of 91.02%. Moreover, we investigate the effectiveness of each stylometric features category with the help of an ablation study. We found that combining all categories of the stylometric features outperforms the other combinations. Finally, we cross compare the feature spaces and classification methods of all solutions. We found that (i) our solution can scale as the number of candidate authors increases, (ii) our method outperforms all the competitors, and (iii) our feature space provides better performance than the feature space used by the existing study.

show abstract

“…Reference [3] used a multi-head recurrent neural network (RNN) as the basic framework. Reference [8] proposed a neural language model based approach. Reference [20] presented a multi-channel convolutional neural network (CNN) model, which integrates character embeddings and word embeddings.…”

Section: Related Workmentioning

confidence: 99%

Multi-Task Learning for Authorship Attribution via Topic Approximation and Competitive Attention

2019

View full text Add to dashboard Cite

Separating content from style is a fundamental problem in authorship attribution to represent topic independent personal style of authors. Previous work often ignores this problem by imposing strong but unrealistic assumptions or artificially determines a set of predefined stylistic structures. This paper proposes to separate topic and style based on neural multi-task learning. Our target is to learn separate representations for topic and style respectively. In addition to authorship attribution as the main task, we introduce a novel auxiliary task topic approximation to guide the learning of topic representations with the topic distributions inferred by topic models, which are trained from external corpus. Moreover, we propose a competitive attention mechanism and a separation-reconstruction constraint to assign different and competitive attentions to two tasks in order to separate topic and style as much as possible. Evaluation results demonstrate that the proposed multi-task learning based method is promising, especially on crosstopic settings. We found that topic approximation can help capture topical content and the competitive attentions benefit topic-style separation. It is encouraging since our model separates topic and style in a probabilistic way and doesn't require human intervention.

show abstract

Domain Specific Author Attribution based on Feedforward Neural Network Language Models

Cited by 12 publications

References 14 publications

Effective writing style transfer via combinatorial paraphrasing

Effective writing style transfer via combinatorial paraphrasing

StyloThai:

Multi-Task Learning for Authorship Attribution via Topic Approximation and Competitive Attention

Contact Info

Product

Resources

About