Detecting Anorexia in Spanish Tweets

Proceedings of Recent Advances in Natural Language Processing, pages 655–663, Varna, Bulgaria, Sep 2–4, 2019. https://doi.org/10.26615/978-954-452-056-4_077 655 Detecting Anorexia in Spanish Tweets Pilar L´opez- ´Ubeda, Flor Miriam Plaza-del-Arco, Manuel Carlos D´ıaz-Galiano, L. Alfonso Ure˜na-L´opez, Maria-Teresa Mart´ın-Valdivia Department of Computer Science, Advanced Studies Center in ICT (CEATIC) Universidad de Ja´en, Campus Las Lagunillas, 23071, Ja´en, Spain {plubeda, fmplaza, mcdiaz, laurena, maite}@ujaen.es Abstract Mental health is one of the main con- cerns of today’s society. Early detection of symptoms can greatly help people with mental disorders. People are using social networks more and more to express emo- tions, sentiments and mental states. Thus, the treatment of this information using NLP technologies can be applied to the au- tomatic detection of mental problems such as eating disorders. However, the first step for solving the problem should be to pro- vide a corpus in order to evaluate our sys- tems. In this paper, we specifically focus on detecting anorexia messages on Twitter. Firstly, we have generated a new corpus of tweets extracted from different accounts including anorexia and non-anorexia mes- sages in Spanish. The corpus is called SAD: Spanish Anorexia Detection corpus. In order to validate the effectiveness of the SAD corpus, we also propose several machine learning approaches for automat- ically detecting anorexia symptoms in the corpus. The good results obtained show that the application of textual classification methods is a promising option for develop- ing this kind of system demonstrating that these tools could be used by profession- als to help in the early detection of mental problems. 1 Introduction Mental health is one of the main concerns of to- day’s society. The World Health Organisation es- timates that 1 in 4 individuals experience men- tal disorders at some stage of their lives. Glob- ally, it is estimated that about 450 million people worldwide are mentally ill, with this kind of ill- ness making up 13% of diseases around the world (Vos et al., 2015). Traditionally, mental health evaluation is based on face-to-face interviews, self-reported issues or the distribution of questionnaires, which is usually labor-intensive and time consuming. However, in recent years several studies have used different technologies to improve the detection of mental health issues. Specifically, some interesting stud- ies explore the relationship between data from on- line social networks and users’ mental conditions (Rahman et al., 2018). Some of them focus on stress (Thelwall, 2017; Lin et al., 2017), depres- sion (Tsugawa et al., 2015), suicide (O’Dea et al., 2015; Astoveza et al., 2018) or anxiety (Shen and Rudzicz, 2017), and most of them use and extract data from Twitter, probably because the informa- tion is open and more accessible than on other platforms, and also because it is one of the most popular social networks among young people. In this paper we focus on mental health problems re- lated to eating disorders because they exhibit the highest mortality rate of any mental illness and 20% of all deaths from anorexia are the result of suicide (Arcelus et al., 2011). …

Content truncated to 500 words