Knowledge Discovery in Portuguese Language Suicidal Tweets Using Filtered- Extended Association Rules Network

The World Health Organization (WHO) and the Global Burden of Disease study estimate that nearly 800,000 people die by suicide each year. Social media are emerging surveillance tools that can help researchers track suicide risk factors in real time. Text Mining naturally becomes an area with greater affinity to promote studies in media such as Twitter. The discovery of terms that imply suicidality becomes crucial for its classification. The present article proposes the use of Filtered-Extended Association Rules Networks, for the selection of terms that indicate or not a suicidal tendency. The results provided the discovery of sets of terms that can be used to help classify tweets as being suicidal or not.