Inferring Gender from the Content of Tweets: A Region Specific Example

Abstract

There is growing interest in using social networking sites such as Twitter to gather real-time data on the reactions and opinions of a region's population, including locations in the developing world where social media has played an important role in recent events, such as the 2011 Arab Spring. However, many interesting and important opinions and reactions may differ significantly within a given region depending on the demographics of the subpopulation, including such categories as gender and ethnicity. Unfortunately, the demographic characteristics of social media users are often unknown because such categories are not always captured in user metadata. Twitter, for example, does not capture a user’s gender in their profile, and inferring gender from first names is difficult since Twitter users are not required to give their real names. There is thus a need for automated methods that can infer such hidden attributes of users from other data sources. In this paper we describe a method to infer the gender of Twitter users from only the content of their tweets. Looking at Twitter users from the West African nation of Nigeria, we applied supervised machine learning using features derived from the content of user tweets to train a classifier. Using unigram features alone, we obtained an accuracy of 80% for predicting gender, suggesting that content alone can be a good predictor of gender. An analysis of the highest weighted features shows some interesting distinctions between men and women both topically and emotionally. We argue that approaches such as the one described here can give us a clearer picture of who is utilizing social media when certain user attributes are unreliable or not available.

Linguistic Bias in Collaboratively Produced Biographies: Crowdsourcing Social Stereotypes?

“…Some scholars have documented the linguistic style of particular media, in order to better understand how users exploit and expand their communication affordances (e.g., (Hu, Talamadupula, and Kambhampati 2013)) or when and why people might change their linguistic style (Danescu-Niculescu-Mizil et al 2013;Michael and Otterbacher 2014). In contrast, others have studied the correlation between participants' linguistic patterns and their offline demographics and identities (e.g., (Pennacchiotti and Popescu 2011;Fink, Kopecky, and Morawski 2012;Park et al 2013).…”

"How Old Do You Think I Am?" A Study of Language and Age in Twitter

“…Experiments on automatic classification of users according to latent attributes such as gender and age have been done on a wide range of resources, including telephone conversations (Garera and Yarowsky 2009), blogs (Sarawgi, Gajulapalli, and Choi 2011), forum posts (Nguyen, Smith, and Rosé 2011) and scientific articles (Bergsma, Post, and Yarowsky 2012;Sarawgi, Gajulapalli, and Choi 2011). Recently, Twitter has started to attract interest by researchers as a resource to study automatic identification of user attributes, such as ethnicity (Pennacchiotti and Popescu 2011;Rao et al 2011), gender (Fink, Kopecky, and Morawski 2012;Bamman, Eisenstein, and Schnoebelen 2012;Rao et al 2010;Burger et al 2011;Rao et al 2011), geographical location (Eisenstein et al 2010) and age (Rao et al 2010).…”

“…Therefore, their set of users was restricted to users having blogs and willing to link them using Twitter. Some approaches used lists of male and female names, for example obtained using Facebook (Fink, Kopecky, and Morawski 2012) or from the US social security department (Zamal, Liu, and Ruths 2012;Bamman, Eisenstein, and Schnoebelen 2012).…”

Detecting Journalism in the Age of Social Media:Three Experiments in Classifying Journalists on Twitter

“…A large body of work has been focused on how to infer users attributes or classify users into certain category according to their explicit or implicit characteristics. Previous work has looked at identifying user's age (Rao et al 2010), gender (Fink, Kopecky, and Morawski 2012), ethnicity (Pennacchiotti and Popescu 2011), regional origin (Huang, Weber, and Vieweg 2014;Rao et al 2010), political affiliation (Cohen and Ruths 2013;Pennacchiotti and Popescu 2011), interest in particular articles (Carreira et al 2004), roles in a conversation (Tinati et al 2012), and etc.…”