The effect of author set size and data size in authorship attribution

Luyckx, Kim; Daelemans, Walter

doi:10.1093/llc/fqq013

Cited by 107 publications

(71 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…found that Document Author Representation(DAR) can be very useful in AA tasks, because it provides good performance on imbalanced data, getting comparable or better accuracy results. [5,6] investigated the class imbalance problem and tests several methods for compensation of imbalanced data sets. He concludes that the best method uses many short text samples for minority classes and less but longer ones for the majority classes.…”

Section: Literature Surveymentioning

confidence: 99%

Authorship Attribution on Imbalanced English Editorial Corpora

Rao¹,

Raju²,

Kumar³

2017

IJCA

View full text Add to dashboard Cite

Authorship attribution is one of the important problem, with many applications of practical use in the real-world. Authorship identification determines the likelihood of a piece of writing produced by a particular author by examining the other writings of that author. Every author has a unique style of writing pattern. This paper identifies the unique style of an author(s) using lexical stylometric features including function words using balanced training corpus. The present paper calculates the frequencies of the lexical based stylometric features by balancing training and test corpus on English editorial documents. The present paper compares various machine learning algorithms for the authorship attribution and achieved highest average accuracy 95.58 using Random Forest classifier and 92.59 using Multilayer Perceptron algorithms.

show abstract

Section: Literature Surveymentioning

confidence: 99%

Authorship Attribution on Imbalanced English Editorial Corpora

Rao¹,

Raju²,

Kumar³

2017

IJCA

View full text Add to dashboard Cite

show abstract

“…Among all these stylometric features, the character n-gram model in character-based linguistic modality performs the best, and it is comparatively more robust against the others [Luyckx and Daelemans 2011;Koppel et al 2011]. The character n-gram model actually captures information crossing different modalities [Houvardas and Stamatatos 2006]; for example, a frequent 'ed' bigram in a character-based modality may also carry the frequent usage of past tense in a syntactic modality.…”

Section: Stylometric Featuresmentioning

confidence: 99%

“…On the contrary, short snippets are relatively casual, and their stylometric features have larger variation. As shown in recent research [Koppel et al 2011;Luyckx and Daelemans 2011;Narayanan et al 2012], authorship attribution accuracy is greatly and directly affected by many objective factors (e.g., text length, number of known author samples, etc.) due to the unstructured nature of the text itself.…”

Section: Introductionmentioning

confidence: 99%

A Visualizable Evidence-Driven Approach for Authorship Attribution

Ding

Fung

Debbabi

2015

ACM Trans. Inf. Syst. Secur.

View full text Add to dashboard Cite

The Internet provides an ideal anonymous channel for concealing computer-mediated malicious activities, as the network-based origins of critical electronic textual evidence (e.g., emails, blogs, forum posts, chat log etc.) can be easily repudiated. Authorship attribution is the study of identifying the actual author of the given anonymous documents based on the text itself, and, for decades, many linguistic stylometry and computational techniques have been extensively studied for this purpose. However, most of the previous research emphasizes promoting the authorship attribution accuracy and few works have been done for the purpose of constructing and visualizing the evidential traits; also, these sophisticated techniques are difficult for cyber investigators or linguistic experts to interpret. In this paper, based on the EEDI (End-to-End Digital Investigation) Framework we propose a visualizable evidence-driven approach, namely VEA, which aims at facilitating the work of cyber investigation. Our comprehensive controlled experiment and stratified experiment on the real-life Enron email data set both demonstrate that our approach can achieve even higher accuracy than traditional methods; meanwhile, its output can be easily visualized and interpreted as evidential traits. In addition to identifying the most plausible author of a given text, our approach also estimates the confidence for the predicted result based on a given identification context and presents visualizable linguistic evidence for each candidate.

show abstract

“…If such a characteristic was absent in an anonymous text, it did not necessarily argue against a writer's authorship in whose other texts (perhaps in different topics or genres) the characteristic did prominently feature. Apart from the limited scalability of such style (Luyckx, 2010;Luyckx and Daelemans, 2011), a far more troublesome issue is associated with them. Because of their whimsical nature these low-frequency phenomena could have struck an author's imitators or followers as strongly as they could have struck a scholar.…”

Section: Seminal Workmentioning

confidence: 99%

Proceedings of the 3rd Workshop on Computational Linguistics for Literature (CLFL)

Feldman¹,

Kazantseva²,

Śzpakowicz³

2014

View full text Add to dashboard Cite

ii Preface Welcome to the third edition of the workshop on Computational Linguistics for Literature. What started out two years ago as a small affair with an unorthodox title seems to be shaping into a modestly sized but vibrant research community.Thanks to the effort of the authors and the program committee, April 27, 2014 promises to be an interesting day. This year's workshop boasts the rich pickings of papers focussed on creating character representations from text and on applications of such representations. Agarwal et al. present a system for extracting social networks from movie scripts. Coll Ardanuy and Sporleder propose a method of finding similar novels using social networks as the underlying representation. Bullard and Alm work on identifying social information about characters from dialogues in a corpus of plays. Taking this a step further, Iosif and Mishra describe an integrated system for identifying and classifying characters in children's stories based on direct and indirect speech.Taking a complementary point of view, the position paper by Levison and Lessard builds upon their previous work and proposes a graph-based representation for the temporal structure of narratives. The paper by Zemánek and Milička describes a diachronic study of Arabic literature, tracing the influence of certain treatises across centuries.In the categories all of their own are two more papers. Davis and Mohammad break new ground with a paper on generating music from literature; you can listen to a one-minute musical summary of a novel. Mike Kestemont presents a position paper on authorship attribution, a topic which was quite popular in the 2012 and 2013 editions of the workshop.Last but by no means least, we are delighted to have two invited speakers this year. An artist and researcher María Mencía from Kingston University in London will talk about electronic poetry. Jan Christoph Meister from University of Hamburg will tell us about the connections between narratology and computation.We look forward to seeing you in Göteborg. Language-Art Digital Poetics: An exploration of digital textualities in the production of artistic research María Mencía Faculty of Arts and Social Sciences Kingston University AbstractAs an academic and an artist my practice-based research is at the intersection of language, art and digital technology. It explores the area of the in-between, the visual, the aural and the semantic. I am always interested in experimenting with the digital medium with the aim of engaging the reader/viewer/user in an experience of shifting 'in' and 'out' of language. This involves looking 'at' and looking 'through' transparent and abstract textualities and linguistic soundscapes. It draws from avant-garde poetics remediating concepts of reading and writing, exploring digital media grammars (voice activation, use of webcam, use of mouse, acts of revealing, triggering, cut and paste, dragging) for interactivity, aesthetics, engagement and meaning production. It is trans-disciplinary, bringing together different cultu...

show abstract

The effect of author set size and data size in authorship attribution

Cited by 107 publications

References 18 publications

Authorship Attribution on Imbalanced English Editorial Corpora

Authorship Attribution on Imbalanced English Editorial Corpora

A Visualizable Evidence-Driven Approach for Authorship Attribution

Proceedings of the 3rd Workshop on Computational Linguistics for Literature (CLFL)

Contact Info

Product

Resources

About