Deep Stylometry and Lexical &amp; Syntactic Features Based Author Attribution on PLoS Digital Repository

Hassan, Saeed‐Ul; Imran, Mubashir; Iftikhar, Tehreem; Safder, Iqra; Shabbir, Mudassir

doi:10.1007/978-3-319-70232-2_10

Cited by 16 publications

(9 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A potential limitation of this study is the adoption of the definitions that as such came with the dataset [4,14]. In future studies, other definitions and features could be explored, such as stylometric features from full-text [15].…”

Section: Discussionmentioning

confidence: 99%

“…Numerous studies discuss the issue of identifying the importance of citations using supervised machine learning techniques applying contextual and quantitative features [3][4][5]. The algorithms and techniques to approach a certain research problem, as well as the writing style of the author [6], contribute greatly in making an article influential. The number of citations received by scientific literature often accounts for their quantitative impact, but not all citations can be considered equal.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Mining the Context of Citations in Scientific Publications

Hassan

Iqbal

Imran

et al. 2018

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Recent advancements in information retrieval systems significantly rely on the context-based features and semantic matching techniques to provide relevant information to users from ever-growing digital libraries. Scientific communities seek to understand the implications of research, its importance and its applicability for future research directions. To mine this information, absolute citations merely fail to measure the importance of scientific literature, as a citation may have a specific context in full text. Thus, a comprehensive contextual understanding of cited references is necessary. For this purpose, numerous techniques have been proposed that tap the power of artificial intelligence models to detect important or incidental (non-important) citations in full text scholarly publications. In this paper, we compare and build upon on four state-of-the-art models that detect important citations using 450 manually annotated citations by experts-randomly selected from 20,527 papers from the Association for Computational Linguistics corpus. Of the total 64 unique features proposed by the four selected state-of-the-art models, the top 29 were chosen using the Extra-Trees classifier. These were then fed it to our supervised machine learning based models: Random Forest (RF) and Support Vector Machine. The RF model outperforms existing selected systems by more than 10%, with 89% precision-recall curve. Finally, we qualitatively assessed important and non-important citations by employing and self-organizing maps. Overall, our research work supports information retrieval algorithms that detect and fetch scientific articles on the basis of both qualitative and quantitative indices in scholarly big data.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Mining the Context of Citations in Scientific Publications

Hassan

Iqbal

Imran

et al. 2018

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

show abstract

“…Boumber et al (2018) propose another CNN approach designed for multilabel attribution tasks, but also take advantage of topic information through word embeddings. Hassan et al (2017) achieve 95% attribution accuracy on scientific papers via a supervised LSTM and lexical and syntactic features. However, since topic seems to not have been controlled during training, it is unclear whether writing style was actually learned.…”

Section: Deep Learning-based Attributionmentioning

confidence: 99%

The Importance of Suppressing Domain Style in Authorship Analysis

Bischoff,

Deckers,

Schliebs

et al. 2020

Preprint

View full text Add to dashboard Cite

The prerequisite of many approaches to authorship analysis is a representation of writing style. But despite decades of research, it still remains unclear to what extent commonly used and widely accepted representations like character trigram frequencies actually represent an author's writing style, in contrast to more domain-specific style components or even topic. We address this shortcoming for the first time in a novel experimental setup of fixed authors but swapped domains between training and testing. With this setup, we reveal that approaches using character trigram features are highly susceptible to favor domain information when applied without attention to domains, suffering drops of up to 55.4 percentage points in classification accuracy under domain swapping. We further propose a new remedy based on domain-adversarial learning and compare it to ones from the literature based on heuristic rules. Both can work well, reducing accuracy losses under domain swapping to 3.6% and 3.9%, respectively.

show abstract

“…uses, downloads, bookmarks, etc.) that contribute useful metrics to the study of research impact apart from traditional citation metrics [5 –7]. The data and indicators on these social networking websites make up the Altmetrics universe [8].…”

Section: Introductionmentioning

confidence: 99%

“…Of these, https://Altmetric.com has the most comprehensive coverage of tweets on research articles and uses this information as an indicator of the articles’ impact [30,31]. Sentiment analysis and opinion mining are undertaken to gauge public opinion of social media posts [32,33], and the interest in using such strategies on news and blogs is growing substantially [34,35]. Since scholars are using social media platforms to share and discuss their research and Twitter is the focus of Altmetrics research, sentiment analysis of tweets about research articles can yield valuable insights into public opinion and the early impact of scientific literature.…”

Section: Introductionmentioning

confidence: 99%

Sentiment analysis of tweets through Altmetrics: A machine learning approach

Hassan

Saleem

Soroya

et al. 2020

Journal of Information Science

Self Cite

View full text Add to dashboard Cite

The purpose of the study is to (a) contribute to annotating an Altmetrics dataset across five disciplines, (b) undertake sentiment analysis using various machine learning and natural language processing–based algorithms, (c) identify the best-performing model and (d) provide a Python library for sentiment analysis of an Altmetrics dataset. First, the researchers gave a set of guidelines to two human annotators familiar with the task of related tweet annotation of scientific literature. They duly labelled the sentiments, achieving an inter-annotator agreement (IAA) of 0.80 (Cohen’s Kappa). Then, the same experiments were run on two versions of the dataset: one with tweets in English and the other with tweets in 23 languages, including English. Using 6388 tweets about 300 papers indexed in Web of Science, the effectiveness of employed machine learning and natural language processing models was measured by comparing with well-known sentiment analysis models, that is, SentiStrength and Sentiment140, as the baseline. It was proved that Support Vector Machine with uni-gram outperformed all the other classifiers and baseline methods employed, with an accuracy of over 85%, followed by Logistic Regression at 83% accuracy and Naïve Bayes at 80%. The precision, recall and F1 scores for Support Vector Machine, Logistic Regression and Naïve Bayes were (0.89, 0.86, 0.86), (0.86, 0.83, 0.80) and (0.85, 0.81, 0.76), respectively.

show abstract

Deep Stylometry and Lexical & Syntactic Features Based Author Attribution on PLoS Digital Repository

Cited by 16 publications

References 11 publications

Mining the Context of Citations in Scientific Publications

Mining the Context of Citations in Scientific Publications

The Importance of Suppressing Domain Style in Authorship Analysis

Sentiment analysis of tweets through Altmetrics: A machine learning approach

Contact Info

Product

Resources

About