Domestic violence (DV) is a serious social problem that can bring many psychological impacts on victims and even witnesses. A survey conducted in 2002 shows that about 30% of China's families exist domestic violence. Research on the impact of domestic violence, especially in the short term, can help us know more about the victims and offer better help to them.  

However, as domestic violence is an emergency, it is difficult to measure the mental status of DV victims before their first experience by using traditional methods. Therefore, we proposed that a new method, Online Ecological Recognition (OER), which can measure the mental health of DV victims shortly before and after DV with dating data on Online Social Networks. 

In the study, researchers first manually identified 232 DV victims from over 1 million users who used DV keywords in their posts.  

Then the mental health prediction modes were used to predict the mental health of the victims before DV (T1) and 4 weeks after DV (T2).  

The predictive models were based on 88 linguistic features and 11 behavioral features extracted from Weibo posts. Predicted mental health includes depression, suicide probability and life satisfaction.  

The victims were divided into separate groups according to the type of DV to provide more information about the effects of DV. The groups are: 1) victims of intimate child abuse, 2) victims of child abuse, 3) victims who exposure to DV. 

Fig. 1. Using the mental health prediction modes to predict the mental health of the victims before DV (T1) and 4 weeks after DV (T2). (Image by LIU Mingming)