Assessing socioeconomic status of Twitter users: A survey

Proceedings of Recent Advances in Natural Language Processing, pages 388–398, Varna, Bulgaria, Sep 2–4, 2019. https://doi.org/10.26615/978-954-452-056-4_046 388 Assessing Socioeconomic Status of Twitter Users: A Survey Dhouha Ghazouani1, Luigi Lancieri2, Habib Ounelli1 and Chaker Jebari3 1Faculty of Sciences of Tunis, University Campus, Tunis 1060, Tunisia (dhouha.ghazouani,habib.ounelli)@fst.utm.tn 2Univ.Lille, Research Center in Signal and Automatic Computing of Lille, F-59000 Lille, France luigi.lancieri@univ-lille.fr 3Information Technology Department, Colleges of Applied Sciences, Ibri, Oman jebarichaker@yahoo.fr Abstract Every day, the emotion and opinion of dif- ferent people across the world are reflected in the form of short messages using mi- croblogging platforms. Despite the exis- tence of enormous potential introduced by this data source, the Twitter community is still ambiguous and is not fully explored yet. While there are a huge number of studies examining the possibilities of in- ferring gender and age, there exist hardly researches on socioeconomic status (SES) inference of Twitter users. As socioeco- nomic status is essential to treating diverse questions linked to human behavior in sev- eral fields (sociology, demography, public health, etc.), we conducted a comprehen- sive literature review of SES studies, in- ference methods, and metrics. With ref- erence to the research on literature’s re- sults, we came to outline the most crit- ical challenges for researchers. To the best of our knowledge, this paper is the first review that introduces the different as- pects of SES inference. Indeed, this article provides the benefits for practitioners who aim to process and explore Twitter SES in- ference. 1 Introduction The ability to identify the socioeconomic sta- tus of social media users accurately is beneficial for the individual scale as well as the societal one. This field starts to be a well-explored research do- main. The difficulty to identify the socioeconmic status of authors and the lack of explicit personal information have brought with them some chal- lenge for computer scientists. Nowadays, Twitter’s monthly active members ex- ceed 300 millions. These members generate over 500 million conversations (tweets) daily 1. These conversations are short text messages including a maximum of 140 characters (recently extended to 280). Indeed, this shortage of characters leads to unstructured and noisy texts to the point that natural language processing (NLP) tools cannot manage successfully (Ritter et al., 2011). More- over, more deduction is required to detect the un- derlying features of Twitter users. In this regard, researchers and specialized centers will explore and analyze the available demographic informa- tion of Twitter users. The results provided by the Pew Research Center (Smith and Brenner, 2012), a subsidiary of the Pew Charitable Trust, show that the majority of Twitter users in the United States are young, with high educational level and exposing a bigger political interest. Authors con- cluded that focusing on a community character- ized by high level of involvement in societal is- sues (Li et al., 2015) could be fruitful. In a first study (Preot¸iuc-Pietro et al., 2015a) prove that lan- guage use in social media is an indicator of user’s occupational class. In a second study, …

Content truncated to 500 words