‘Violence’ Plus ‘Depression’ Is ‘Parents’: Text Mining Analysis on Korean Student Suicide Attempt Case Reports with Word2vec
A student suicide attempt case report is an official document that is submitted to Korean ministry of education if any suicide attempt case occurs. The report is made on basis of a teacher’s direct or indirect observations (counselling with the suicide attempted student or his/her parents). The report contains valuable information such as family and peer relationship, academic achievement, and psychiatric symptoms of the students. However, the information is depicted in linguistically written and ill-organized form, so it is difficult to extract patterns or tendencies of Korean students’ suicide attempt. A machine learning technique called ‘word2vec’ is applied to extract patterns of the report.The word2vec technique allows that terms in the report are represented in a continuous vector space where semantically similar words are mapped to nearby points. Throughout the text mining analytics of the reports, the problematic words such as ‘violence’, ‘depression’ and ‘child abuse’ are strongly associated with family members – father, mother or both of them (parents). Furthermore, these problematic words are placed nearby in the vector space, which means that problems affect students’ suicide attempt are usually accompanied together, not isolated. The implication of our results will be discussed in terms of the school-based prevention program.
Keywords - Student Suicide Attempt Case Report, Word2vec, Vector Space,