Internet Suicide Notes Recognition Research

With the development of science and technology, people's living standards have been improved, but also accelerated by the pace of life, modern people face tremendous pressure. Driven by various pressures, the phenomenon of suicide has been intensified in recent years, and affected the country's economic and social development seriously. Some people who committed suicide left behind a suicide note before committing suicide, and the contents of the suicide note reflect the reasons for suicide. Countries around the world are looking for effective ways to prevent people suicide, but still did not form a complete set of well-established mechanism.With the development of the Internet, the way people live has undergone enormous changes, and the suicide note from the original "paper suicide note" into "electronic suicide note" which appeared in recent years, with the rise of micro blog and the forum which has been prevalent. The purpose of this research is how to identify those web texts with suicidal tendencies on the micro blog and forum platform.This thesis analyzes a variety of academic literature in this field at home and abroad firstly, and then tries to use a new and more effective method to identify the Internet suicide note. We collected a sample of the suicide notes that needed in this research, and verified their authenticity according to the list of keywords provided by CCNU's Psychological College Youth Cyber Psychology and Behavior, the Ministry of Education Key Laboratory for screening, and added a number of discriminating vocabularies. On the other hand, considering the suicide note has very unique characteristics:it describes a very specific fact, but the small differences between the suicide notes but large gap between species. The problem that using text classification algorithms to identify the problems faced is that despite the false detection rate will be lower, but the detection rate is not very satisfactory too. To solve the problem it raises, a approach of combining the keyword scanning algorithm and text classification algorithm is designed:firstly the test sample will be identified by the text classification algorithm. If considered to have suicidal tendencies, then directly as it is, otherwise using keyword scanning algorithm further to make a decision. Experimental result shows that the algorithm is effective.