A visual approach for age and gender identification on Twitter

Abstract

The goal of Author Profiling (AP) is to identify demographic aspects (e.g., age, gender) from a given set of authors by analyzing their written texts. Recently, the AP task has gained interest in many problems related to computer forensics, psychology, marketing, but specially in those related with social media exploitation. As known, social media data is shared through a wide range of modalities (e.g., text, images and audio), representing valuable information to be exploited for extracting valuable insights from users. Nevertheless, most of the current work in AP using social media data has been devoted to analyze textual information only, and there are very few works that have started exploring the gender identification using visual information. Contrastingly, this paper focuses in exploiting the visual modality to perform both age and gender identification in social media, specifically in Twitter. Our goal is to evaluate the pertinence of using visual information in solving the AP task. Accordingly, we have extended the Twitter corpus from PAN 2014, incorporating posted images from all the users, making a distinction between tweeted and retweeted images. Performed experiments provide interesting evidence on the usefulness of visual information in comparison with traditional textual representations for the AP task.

Born digital or fossilised digitally? How born digital data systems continue the legacy of social violence towards LGBTQI + communities: a case study of experiences in the Republic of Ireland

“…Gender identity and gender classification are also of particular interest to law enforcement for forensics, investigation and social justice and to commercial organizations for marketing and advertising and to others for social reasons (Alowibdi et al 2013;Álvarez-Carmona et al 2018). As a result, there have been numerous studies using data mining techniques to identify the gender of users on social media sites such as Twitter, YouTube and Flickr (Burger et al 2011;Cheng et al 2011;Peersman et al 2011;Alowibdi et al 2013).…”

“…"Experiments also indicate that function words, word-based features and structural features are significant gender discriminators" (Cheng et al 2011). However, all these studies have defined gender identity as a binary classification problem consisting of two classes, 'male' and 'female' (Burger et al 2011;Cheng et al 2011;Peersman et al 2011;Alowibdi et al 2013;Eltaher and Lee 2015;Álvarez-Carmona et al 2018). Shoshana Zuboff (2019) expresses the collection of online user data for translation into behavioural data for consumption by machine learning techniques and algorithms to predict our future behaviour as 'surveillance capitalism', "an information civilization shaped by surveillance capitalis and its new instrumentarian power will thrive at the expense of human nature and will threaten to cost us our humanity" (Zuboff 2019, p 11-12).…”

Gender Identification from Turkish Tweets Using Pre-Trained Language Models

“…Eğitim aşamasında başarımı artırabilmek için aynı yazar tarafından gönderilen metinler N boyutunda bir çerçeve kullanılarak birleştirilmiştir. n= [1,2,4,5,10] olmak üzere 5 farklı değer seçilmiştir. Yazara ait yazım tarzını modelin öğrenebilmesi için çerçeve veri seti üzerinde kaydırılarak veri seti her defasında yeniden düzenlenmiştir.…”

“…Yaş, cinsiyet, eğitim gibi temel özelliklerin tespiti görevlerinin yanında, herhangi bir yazarın sahte haber yaymaya istekli olup olmadığının belirlenmesi gibi özel görevler de bulunmaktadır [1]. PAN2013 [2,3] ten itibaren bu görevler içerisine giren cinsiyet tespiti bugün bile önemini korumaktadır. Özellikle siber suçların tespitinde önemli bir yeri olan cinsiyet belirleme görevi İngilizce, İspanyolca gibi dillerde önemli başarılara ulaşmasına rağmen Türkçe için yeterli sayıda çalışma yapılmamıştır.…”

An Author Gender Detection Method Using Whale Optimization Algorithm and Artificial Neural Network

“…They found out that SVM and deep learning were the most popular and successful classifiers in author profiling. Alvarez-Carmona et al [19] evaluated author profiling detection based on textual and visual resources on Twitter to recognize some demographic aspects such as age and gender. The authors applied a dynamic feature selection on images posted on Twitter.…”