An open stylometric system based on multilevel text analysis

Eder, Maciej; Piasecki, Maciej; Walkowiak, Tomasz

doi:10.11649/cs.1430

Cited by 16 publications

(17 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In stylometric authorship studies, researchers have generally used various statistical multivariate analysis techniques that range from frequency distribution (i.e. listing frequently used words) to discriminant analysis examining linguistic and stylistic features within texts that can be detectors of potential authors (Daelemans, 2013;Eder, Piasecki & Walkowiak, 2017;Gómez-Adorno, Posadas-Duran, Ríos-Toledo, Sidorov & Sierra, 2018;Hoover, 2003;Lagutina, Boychuk, Vorontsova & Paramonov, 2019). These statistical techniques have been used with different linguistic variables, including morphological, lexical, and syntactic variables (Burrows, 2007;Strome, 2013).…”

Section: Literature Reviewmentioning

confidence: 99%

Authorship attribution of Morsi Gameel Aziz’s lyrics: A clustering-based stylometry approach

Omar

2021

JLLS

View full text Add to dashboard Cite

Numerous studies have addressed the issue of the authorship of Morsi Gameel Aziz's lyrics. These studies have traditionally been based on chronological criteria for determining the real authors of disputed lyrics. To date, there is no agreement on the real authors of these disputed lyrics. This can mainly be attributed to both selectivity and the lack of empirical evidence in such studies, raising questions about the reliability of such approaches. With the advent of machine learning systems and data mining techniques, it is now possible to process thousands of texts using replicable methods. Thus, this study seeks to address the issue of the authorship of Morsi Gameel Aziz's lyrics making use of these advances by applying a clustering-based stylometry approach. The hypothesis is that lyrics grouped or clustered together are more likely to be written by the same poet. A corpus of 1,089 lyrics was built, including all known lyrics attributed to Aziz and the lyrics of the poets thought to be the real authors of the disputed lyrics. The lyrics were clustered using the Gibbs sampling Dirichlet multinomial mixture (GSDMM) technique and were assigned to four main classes, with the 12 disputed lyrics clustered within Aziz's class. Based on this, it is clear that the GSDMM model is effective and reliable in clustering short documents in Arabic. The results of the study show that machine learning systems and stylometric authorship techniques can be used in resolving many authorship questions that remain controversial and unanswered in Arabic literature.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

Authorship attribution of Morsi Gameel Aziz’s lyrics: A clustering-based stylometry approach

Omar

2021

JLLS

View full text Add to dashboard Cite

show abstract

“…Stilometrija, statistična ali kvantitativna analiza stila, raziskuje po dobnosti in razlike med besedili na različnih jezikovnih ravninah (Eder et al 2017: 1), v prvi vrsti z namenom določanja avtorstva. Izhodišče stilometrije je, da je avtorjev stil nekaj nezavednega, česar ni mogoče zavestno usmerjati, obenem pa ima kvantitativne razlikovalne lastnosti.…”

Section: (Računalniško Podprta) Stilometrijaunclassified

“…Ko govorimo o stilometriji v okvirih digitalne humanistike, seveda govorimo o računalniško podprti stilometriji; kot taka je definirana kot del računalniških analiz besedil v okviru oddaljenega branja ali makroanalize, in sicer z upoštevanjem velikih (večjih) zbirk besedil, v katerih skuša najti razmerja in vzorce podobnosti in razlik, ki so »očesu človeškega bralca« skriti (Eder et al 2016: 108). Običajno se stilometrija opira na enostavne jezikovne značilnosti, ki jih lahko v besedilnem dokumentu določamo avtomatsko, kot na primer relativna frekventnost (najpogostejših) besed, raba in razporeditev ločil, povprečna dolžina povedi ali besed (Eder et al 2017: 1−2). Še vedno močno prednjači analiza na ravni leksike, vedno večja zmogljivost računalnikov in dostopnost besedil v elektronski obliki pa omo goča ta tudi skladenjske in pomenske stilometrične analize (Holmes, Kardos 2003: 6).…”

Section: (Računalniško Podprta) Stilometrijaunclassified

“…izkušenih programerjev, torej strokovnjakov s tehničnim ozadjem, kot tudi raziskovalcev v humanistiki, ki v računalništvu niso podkovani. Zato so bila razvita računalniška orodja, ki naj bi premostila razkol med obema disciplinama in so za uporabo dostopna raziskovalcem v humanistiki (Eder et al 2017: 1−2), eno od teh orodij je paket Stylo v programskem okolju R, ki ga razvija skupina raziskovalcev s področja jezikoslovja, literarnih ved in računalništva z različnih raziskovalnih ustanov na Poljskem 2 in v Belgiji, Computational Stylistics Group. 3…”

Section: Interdisciplinarnost Stilometrijeunclassified

See 1 more Smart Citation

Računalniško podprta stilometrična analiza pripovedne literature Janeza Ciglerja in Christopha Schmida v slovenščini

Žejn

2020

Fluminensia (Online)

View full text Add to dashboard Cite

V tridesetih letih 19. stoletja je nastala prva (izvirna) slovenska pripoved Sreča v nesreči (1836) avtorja Janeza Ciglerja. Raziskovalci slovenske literature so velik vpliv na nastanek tega dela pripisovali bavarskemu pisatelju Christophu Schmidu, tako v vsebini kot strukturi, ideji in pripovednemu slogu, pa tudi sledenju duhu časa in upoštevanju tradicije. Podrobneje so se posvečali zlasti vsebinskim primerjavam, ne pa tudi analizam podobnosti stila. S paketom Stylo v statističnem programskem okolju R za napredne stilistične analize zbirk besedil smo zato izvedli stilometrično analizo Ciglerjevih in Schmidovih prevedenih del ter nekaterih prevodov pripovedne literature obravnavanega obdobja. Ob upoštevanju določenih omejitev, kot so omejena količina besedil, prevedena in izvirna literatura in žanrska raznolikost, je analiza potrdila izviren Ciglerjev slog in odprla nova vprašanja glede avtorstva nekaterih del.

show abstract

“…That opens possibilities for automatic categorisation of text documents in terms of the subject areas in any digital collection of documents. It is also an important problem for researchers from different areas of the humanities and social science (Eder et al, 2017).…”

Section: Introductionmentioning

confidence: 99%

Evaluation of vector embedding models in clustering of text documents

Walkowiak¹,

Gniewkowski²

2019

Proceedings - Natural Language Processing in a Deep Learning World

Self Cite

View full text Add to dashboard Cite

The paper presents an evaluation of word embedding models in clustering of texts in the Polish language. Authors verified six different embedding models, starting from widely used word2vec, across fast-Text with character n-grams embedding, to deep learning-based ELMo and BERT. Moreover, four standardisation methods, three distance measures and four clustering methods were evaluated. The analysis was performed on two corpora of texts in Polish classified into subjects. The Adjusted Mutual Information (AMI) metric was used to verify the quality of clustering results. The performed experiments show that Skipgram models with n-grams character embedding, built on KGR10 corpus and provided by Clarin-PL, outperforms other publicly available models for Polish. Moreover, presented results suggest that Yeo-Johnson transformation for document vectors standardisation and Agglomerative Clustering with a cosine distance should be used for grouping of text documents.

show abstract

An open stylometric system based on multilevel text analysis

Cited by 16 publications

References 31 publications

Authorship attribution of Morsi Gameel Aziz’s lyrics: A clustering-based stylometry approach

Authorship attribution of Morsi Gameel Aziz’s lyrics: A clustering-based stylometry approach

Računalniško podprta stilometrična analiza pripovedne literature Janeza Ciglerja in Christopha Schmida v slovenščini

Evaluation of vector embedding models in clustering of text documents

Contact Info

Product

Resources

About